MARTENSITIC TRANSFORMATIONS IN HIGH MAGNETIC FIELDS

advertisement
MARTENSITIC TRANSFORMATIONS IN HIGH MAGNETIC FIELDS
by
MICHAEL KARL KORENKO
B.S., Case Institute of Technology
(1966)
M.S., Case Western Reserve University
(1969)
Submitted in partial fulfillment of the requirements
for the degree of
DOCTOR OF SCIENCE
at the
Massachusetts Institute of Technology
February, 1973
Signature of Author .....
Departtment of Metall urgy and Materials Science
Jan~ry 17, 1973
1
/2*
Certified by ........
Thesis
..........
Accepted by ...... .. .... ......
Supervisor..
.Thesis
Supervisor
.......................
Chairman, Departmental Commitiee on Graduate Students
Archives
MAR 20 1973
i-t**AARt9s5
ABSTRACT
MARTENSITIC TRANSFORMATIONS IN HIGH MAGNETIC FIELDS
by
MICHAEL KARL KORENKO
Submitted to the Department of Metallurgy and Materials Science on
January 17, 1973 in partial fulfillment of the requirements for the
degree of Doctor of Science.
The influence of high magnetic fields on the nucleation,
kinetics, and morphology of the martensitic transformation was investigated with fields up to 140 kOe, using electrical resistivity measurements and optical metallography. Four iron-nickel and iron-nickelmanganese alloys were chosen to cover the range of morphologies from
lath (packet) to plate (lenticular) and the range of kinetics from isothermal C-curve behavior to bursting phenomena.
Magnetic fields accelerate the martensitic transformation in
iron-nickel alloys both by increasing the Gibbs free-energy difference
between the product and parent phases and by allowing the reaction to
take place at higher temperatures. For example, the Fe22.5Ni4.OMn alloy,
which will not transform perceptibily at any temperature without a magnetic field, displays isothermal C-curve kinetics with fields in the
range of 60 to 140 kOe. Thus, employing magnetic fields to induce the
martensitic transformation is very different from lowering the temperature.
Lath and plate martensites in iron-nickel alloys are two extremes of a spectrum of morphologies each of which, in themselves, appear
to be intermediate transition morphologies. The morphological transition
from lath-to-plate is not a simple function of the temperature, Gibbs
free-energy change, nickel content, activation energy, or the temperature
difference between the Curie point and the martensitic-start temperature.
The best kinetic correlation with the morphology in binary iron-nickel
alloys is the rate of transformation which, in turn, can be approximated
by the incubation time or the activation energy divided by the temperature.
Lower temperatures, at constant driving force or at constant activation
energy, favors lath martensite which, up to the present use of magnetic
fields, has always been overshadowed by the large increases in chemical
driving force with decreasing temperature which favors plate martensite.
The average plate volume of the iron-nickel-manganese is
insensitive to temperature or magnetic field strength; however, it is
a strong function of the extent of transformation up to about 7 percent
martensite, and becomes relatively insensitive thereafter. In addition,
the orientation of the martensitic plates and their radius-to-semithickness
ratios are not significantly dependent on the direction of the applied
field.
The quantitative influence of magnetic fields on the martensitic
kinetic behavior imposes a severe test on the current nucleation theories.
The Cohen-Raghavan model can be fitted to much of the data, but there are
important details that remain unexplained. In addition, high-temperature
magnetic annealing and recrystallization experiements in the austenitic
range do not support the hypothesis that the preexisting martensitic
embryos are ferromagnetic in nature.
There is evidence that autocatalytic nucleation sites in the
iron-nickel alloys may be an integral part of the moving austenitemartensite interface. Three important autocatalytic sites are identified:
midrib impingement of a martensitic plate on another plate during thickening of the latter; the transition zone within a martensitic plate
where the lattice-invariant mode switches from twinning to slip; and
grain-boundary impingement by an advancing martensitic plate.
Magnetic fields are shown to be a valuable tool for studying
the martensitic transformation in iron-base alloys. It permits a separation of the effects of driving force and temperature, and provides a
critical test for the current nucleation theories.
Thesis Supervisor:
Title:
Morris Cohen
Ford Professor of Materials
Science and Engineering
TABLE OF CONTENTS
Page
Number
TITLE PAGE.....
ABSTRACT ........
Th..ermodynamic
am...*ics
o.
inMt....i.cF..
M.ae
.
iels..
.............
i..
n. Ma...gn.e.t.i...
..........
1
2
TABLE OF CONTENTS .. ...... .. .. ......
.
.. ... .. ..........
. . ..
4
LIST OF FIGURES
*...........
......
7
LIST OF TABLES
.L.E...00U.......
K.e
ACKNOWLEDGEMENTS
I INTRODUCTION
.or
.
u
.
t..
ic
s
h
.
.e
ati.o..n
.
.
.
.........
....
.
..
.
.
...
...
..... .
.
.....
*........
...
...
10
........
.....
11
.............
. o .....................
..
12
.......
I-1
Classical Thermodynamics in Magnetic Fields .............
12
I-2
Thermodyn amics of Martensite in Magnetic o.....
Fields ..
15
I-3
Advantage s of Magnetic Fields for Research on Martensitic
Transform ations .................. ........ ...
.. ...
..... 23
......
24
Martens itic Transformation Characteristics ...........
24
II DISCUSSION OF LITERATURE ......................
II-1
11-2
........
II-1.1
Kinetics ....................................
24
II-1.2
Morphology ......................
30
II-1.3
Nucleation ....................................
............
34
Magneti cally-Induced Martensitic Transformations ......
41
11-2.1
General Magnetic Transformation Characteristics
41
11-2.2
Pulsed Versus Steady Fields ..................
44
11-2.3
Isothermal Transformation Studies ..............
47
11-2.4
Magnetic Effects on the Microstructure ........
48
11-2.5
Theoretical Treatment to Date .................
50
11-2.5.1
50
.
Thermodynamic Predictions ..........
TABLE OF CONTENTS (continued)
Page
Number
II-2.5.2
11-2.6
III
Kinetic Predictions ..................
Potential Applications ...
.. ...
.....
PURPOSE OF THESIS AND OUTLINE OF WORK ....
..... .......
... .....
.....
..
IV MATERIALS AND PROCEDURES ..................
...
.....
....
...
... e..
.......
e..
..
IV-1
Specimen Preparation and Composition
IV-2
Resistance Measurements .............
.. ...
l
IV-3
Temperature and Magnetic-Field Contro 1
IV-4
Magnetization Measurements ..........
.....
•
...
.....
...
e.
..
IV-5
Optical Microscopy ...............
.....
.......
..
IV-6
Quantitative Metallography ..........
..
..
IV-7
IV-6.1
Grain-Size Measurements .....
IV-6.2
Volume-Fraction Measurements
IV-6.3
Measurement of Mean Volume pe
to-Semithickness Ratio ...... ..
...
..
O...
.......
..
.......
and Radiusate
Measurement of Nucleation Rate ...... .. ..
.....
.......
.....
.......
.....
.......
.....
..
.....
.......
.....
.......
.....
.......
.....
.......
o....o
.......
.....
.......
...
ate
IV-8
Averaging Techniques ................
...
...
...
..
V RESULTS AND DISCUSSION ..................... .. Pl
e....
...
V-1
Material Characteristics ..............
...
V-l.1
Shear Modulus and Density ...... ..
...
V-1.2
Magnetic Properties ............
...
V-1.3
V-2
Chemical Driving Force .........
Fe 28.7 Ni --
V-2.1
(Isothermal) ...........
Kinetics .......................
52
TABLE OF CONTENTS (continued)
Page
Number
V-2.2
V-3
Fe 29.6 Ni -- (Isothermal
V-4
VII
Burs ting) ........
..
o
78
....
81
Kinetics ................
....
81
V-3.2
Morphological Transition Behavior .....
....
85
V-3.3
Activation-Energy Correla tions ........
....
93
V-3.4
Fit with the KCR Model ...
(Bursting)
.............
........
....
101
....
104
V-4.1
Kinetics .................
....
104
V-4.2
Morphology and Autocatalyt ic Nucleation
....
108
....
115
Fe 22.5 Ni 4.0 Mn --
V-6
-*
.
V-3.1
Fe 30.8 Ni --
V-5
VI
Morphology ..............
(Isothermal C-curves) ...
V-5.1
Kinetics .................
115
V-5.2
Morphology and Quantitativ e Metallograp hy
119
V-5.3
Fit with the KCR Model ...
130
Embryo Experiments ..............
135
V-6.1
High-Temperature Magnetic Annealing
135
V-6.2
Low-Temperature Magnetic Annealing and Cyc
Experiments ..............
CONCLUSIONS .........................
SUGGESTIONS FOR FUTURE WORK ........
.
. a
.
. .
. a
.............
ng
... i
e
141
....
•
145
....
•
148
APPENDICES
I
II
Magnetic Clausius -- Clapeyron Equation ...........
149
Quantitative Metallography of Thin Oblate Spheroids
151
BIBLIOGRAPHY .........................................
156
BIOGRAPHICAL NOTE .......................................
165
LIST OF FIGURES
Figure
Number
1
Page
Number
Schematic magnetization curve of an ellipsoidal single
crystal of a paramagnetic material illustrating the
energy changes involved ................................
16
2
Schematic survey of past pulsed-field work ..............
42
3
Schematic of cryogenic temperature controller used during
transformation in applied fields .......................
61
4
Room-temperature magnetization curves ...................
71
5
Chemical free-energy change for the austenite-tomartensite transformation from Equation 27 ..............
74
Transformation Kinetics of Fe 28.7 Ni under zero-field
conditions ........................................
76
Field-induced transformation kinetics of Fe 28.7 Ni at
24.40 C ............. ...................................
77
Lath-like morphology of Fe 28.7 Ni transformed in a field
of 60 kOe at 24.4°C ....................................
79
Transformation kinetics of Fe 29.6 Ni under zero-field
conditions ......................................
82
Field-induced transformation kinetics of Fe 29.6 Ni at
- 20C ......................................
.............
83
6
7
8
9
10
11
12
13
Field-induced transformation kinetics of Fe 29.6 Ni at
+ 90 C ..................................................
84
Morphological transition in Fe 29.6 Ni under zero-field
conditions
(a) Lath-like at - 130 C
(b)Mixture of plate and lath-like martensite at - 196 0C
86
Field-induced morphological transition in Fe 29.6 Ni at
- 20C
(a)Lath-like at 40 kOe
(b)
Mixture
of
plate
and
lath-like
martensite
at
140
kOe
LIST OF FIGURES (continued)
Figure
Number
14
15
16
17
18
Page
Number
An example of a twin-related loose-packet in Fe 29.6 Ni
transformed in field of 60 kOe at - 2°C .................
90
An example of the transition plate morphology in Fe 29.6
Ni transformed in field of 140 kOe at - 20C .............
91
Activation energy versus total Gibbs free-energy change
for Fe 29.6 Ni .......................................
96
Activation energy divided by the temperature versus total
Gibbs free-energy change for Fe 29.6 Ni .................
98
Logarithm of the rate of transformation versus experimental
AWa/T and calculated AU*/T illustrating the morphological
transition in Fe 29.6 Ni ...............................
99
19
Transformation kinetics of Fe 30.8 Ni under zero-field
conditions ........................................
105
20
Field-induced transformation kinetics of Fe 30.8 Ni at
- 200 C ...............................................
107
21
Zero-field bursting morphology of Fe 30.8 Ni
(a) - 46%C
22
23
24
25
26
27
(b) - 1960C .......................................
109
Field-induced structure of Fe 30.8 Ni transformed in a
90 kOe field at - 200C .................................
110
An example of autocatalytic plates emitting from a midrib
in Fe 30.8 Ni transformed under zero-field conditions at
- 460C ......................
.........
...................
112
Field-induced transformation kinetics of Fe 22.5 Ni 4.0
Mn at - 810C .......................................
116
Field-induced transformation kinetics of Fe 22.5 Ni 4.0
Mn at 140 kOe as function of various testing temperatures
117
Time to transform to 0.3% martensite as function of
magnetic field and temperature .........................
118
Morphology of the Fe 22.5 Ni 4.0 Mn alloy transformed in
140 kOe field at various temperatures ...................
120
LIST OF FIGURES (continued)
Figure
Number
28
29
30
Page
Number
Average plate volume for Fe 22.5 Ni 4.0 Mn as function of
percent martensite, temperature, and magnetic-field
intensity ........................................
123
Radius-to-semithickness ratio for Fe 22.5 Ni 4.0 Mn as
function of percent martensite, temperature, and magneticfield intensity .......................................
125
Orientation dependence of the morphology of Fe 22.5 Ni 4.0
Mn as function of direction of applied field (tested at
- 810 C, 140 kOe)
(a)Field perpendicular to specimen axis
(b)Field parallel to specimen axis ......................
31
127
Distribution of the number of plates aligned in a specific
direction and their radius-to-semithickness ratios as a
function of the direction of the applied magnetic field ..
128
Comparison of the KCR Model to the activation energy versus
driving-force data .......................................
132
Comparison of the KCR Model to the AWa/T versus drivingforce data ................................................
133
34
Thermal-magnetic history of a typical embryo experiment ...
136
35
Equilibrium diagram illustrating the Curie temperatures and
and martensitic-transformation range of FeNi alloys .......
138
32
33
36
Low-temperature transformation in Fe 29.6 Ni interrupted by
upquenching, with and without a magnetic field ............ 142
10
LIST OF TABLES
Table
Number
Page
Number
1
Relative Magnitudes of Various Magnetic Energies .........
22
2
Composition of Alloys in Weight Percentages ..............
57
3
Material Characteristics of the Alloys Under Investigation
68
4
5
Incubation Times, Activation Energies, and Driving Forces
for Fe 29.6 Ni .......................................
95
Incubation Times, Activation Energies, and Driving Forces
for Fe 22.5 Ni 4.0 Mn ...................................
131
ACKNOWLEDGEMENTS
The author expresses his sincere appreciation to the following
people for their assistance in conducting the work reported in this
thesis:
Professor Morris Cohen for giving me a free hand to pursue independent research and providing continued guidance over the critical
hurdles;
Miss Marguerite Meyer for her many special favors and general
council;
Mrs. Miriam Rich, Mrs. Jane Operacz, and Mr. Robert Cava for
their invaluable assistance with the optical and quantitative metallography;
Dr. Gilbert Speich for his measurements of the density and shear
moduli of our alloys;
Mr. Larry Rubin, Miss Jean Morrison, and all the other helpful
people at the Francis Bitter National Magnet Lab who made my work
there a valuable and pleasant experience;
Ford Scientific Laboratory and the U. S. Steel Laboratory for
Fundamental Research who supplied the alloys used in this investigation; and the National Science Foundation for supporting this work.
I am particularly indebted to my wife, Elaine, who by contributing to my financial support and by sharing the same interests has
made my stay in the Cambridge/Boston area a truly memorable experience.
I. INTRODUCTION
Intypical steels, the martensitic phase isferromagnetic while the
austenite isparamagnetic. Under these conditions, the application of a
magnetic field can induce an additional free-energy difference between
the two phases.
Inthis thesis, we have utilized this effect in order to
study the nucleation and kinetics of the austenite-to-martensite transformation.
I-1 Classical Thermodynamics inMagnetic Fields
The state of the material under investigation is annealed polycrystalline austenite which is textured to some extent because of the
swaging and recrystallization steps necessary to obtain the specimen size.
In addition, each grain contains several domains arranged in such a way as
to give an overall value of magnetization near zero. It is impossible to
rigorously describe the exact thermodynamic state of this system in a magnetic field. Fortunately, however, most of the complicating factors can
be shown to be negligible, especially under high-field conditions.
The following general relationship among the three magnetic vectors
will serve to define the important magnetic quantities:
i = voi + o
074r1
(1)
where I isthe magnetic induction (ingauss), po is the permeability of a
vacuum ( = 1 gauss/oersted), l isthe magnetic field (inoersted), and a is
the magnetization or magnetic moment per unit volume (inergs/gauss cm3)
A isthe only material parameter inthis equation and is the quantity to
be examined when predicting the effect of an applied magnetic field on an
alloy.
The following discussion applies particularly well to diamagnetic or
paramagnetic materials, but should be used with caution for ferromagnetic
systems; in general, the latter exhibit hysteresis effects and many other
complications which we will consider shortly.
Thermodynamically (1-4) the introduction of a field, A, introduces
another work term to the internal energy, much like the pressure introduces the PV term. When a specimen is placed in a,solenoid, its dipoles
ali'gn with the field, thereby increasing its magnetization. This increases
.the internal energy of the material.
The battery driving the current in
the solenoid supplies the extra energy by working against the back emf
generated by the change in magnetization. The exact expression for the
magnetic work that must be added to the internal energy can be broken into
two terms:
dWmag = +r(•t.d)dV
= d(7-fH 2 dV) + olf(4.dA)dV
(2)
(3)
The first term on the right is just the energy contained in the field of
the empty solenoid and has no material significance. The usual procedure
is to redefine the zero of energy as including the energy of the field in
a vacuum. Since we will be dealing with energy differences, this is of
little consequence.
To put this equation into a useable form, it is
necessary to assume that the magnetization is independent of position
within the domain.
This assumption applies quite well to a completely
homogeneous specimen which is ellipsoidal in shape; however, it may be
violated by material inhomogeneities or demagnetization effects at the
With these assumptions:
boundaries.
dW' mag = PoR.dt
(4)
where I is the total magnetic dipole moment of the system ( = fMdV = MV).
The definition of the internal energy is now:
dU' = TdS - PdV + EidN. + ~0 .- dt
(5)
The extensive parameter descriptive of the magnetic system is the component of the total magnetic moment parallel to the external field, while
the intensive parameter in the energy representation is voH. (2)
The internal energy of a material is raised by a magnetic field,
yet we know that paramagnetic or ferromagnetic materials are attracted to
a magnet.
Why would it seem to go to a higher energy state? The answer
to this dilemma is that the potential energy of the material is lowered
or, alternatively, the overall energy change of the specimen plus the
constant field solenoid is negative.
Physically, we would like to consider the specimen as the system
and to treat the solenoid as a magnetic-field reservoir. We are transforming austenite under constant temperature, pressure, and external
magnetic field.
The correct thermodynamic potential for these conditions
is the partial Legendre transform of the internal energy with respect to
T, P, and H:
U[T,P,H] = U - TS + PV - v 0 to
t1
(6)
dU[T,P,H] = - SdT + VdP + l.PjdN. - Ipof.d
(7)
dU[T,P,H] = dGo + dG
(8)
magnetic
These equations show that we can add the magnetic contribution directly
onto the Gibbs free-energy change.
Figure
(1) illustrates how the magnetic term may be calculated and
offers some physical insight into the relevant thermodynamics. The shaded
area above the magnetization curve isequal to the increase of internal
energy of the specimen ( = po
0 fidt ).
PoHI is the total area of the
dashed rectangle and is equal to the decrease inenergy of the magneticfield reservoir. Finally, oft-.di isthe area under the magnetization
curve and isequal to the energy change of the system plus the reservoir.
When measured under constant temperature and pressure conditions, this
is the magnetic contribution to the Gibbs free energy.
I-2 Thermodynamics of Martensite in Magnetic Fields
The above analysis can be applied to ferromagnetic materials
within a single ellipsoidal domain. However, even inthis idealized
situation, demagnetization and anisotropy effects must be taken into
account. The best way to sort out the additional energy contributions
in ferromagnetic system is to consider each energy term individually(5 -11
1. The exchange energy arises from the quantum-mechanical
interaction of the spins of unpaired electrons.
The "molecular field"
isessentially defined such that the energy of interaction of the magnetic moment of the atom with the molecular field is equal to the exchange
energy. This energy does not enter the thermodynamic picture except as a
---
1=
LOf I -dH
H
-
-
II,
.. C
T,
n
-u,
1 ='F
H=O, T=O
H >
FIG. I
I
IT=T,
"
r
SCHEMATIC MAGNETIZATION CURVE OF AN
ELLIPSOIDAL SINGLE CRYSTAL OF A
PARAMAGNETIC MATERIAL ILLUSTRATING
THE ENERGY CHANGES INVOLVED.
contribution to domain wall energy because we have considered the
domained, ferromagnetic crystal as the standard state, i.e. this
energy has already been included in the standard Gibbs free-energy
change.
The size of the molecular field is estimated to be about 107
gauss.
This is important because it shows that even our "high" applied
fields of 105 gauss will have little effect on the saturation magnetization, Is , which already exists in the domains.
Since I s does not
change significantly with the applied field, dl = 0 and equation (5)
shows that the internal energy due to magnetization will not change with
the field. The mutual energy between the specimen and the field,
dGmag of equation (8), is now equal to - vjOs. d
and is the overriding
factor, especially under high-field conditions.
2. The anisotropy energy or the magnetocrystalline energy
arises because the magnetic dipoles tend to align along certain easy
directions of magnetization. The orbital magnetic moment of the
electrons is strongly coupled to certain lattice directions due to
electrostatic interactions.
The spin magnetic moment of the electrons,
which accounts for more than 90% of the observed magnetization, is
weakly coupled to the orbital moment. Under the influence of an external field, the spin dipoles rotate out of this coupling "easy" direction
in order to reduce the Gibbs magnetic energy. The order of magnitude
of this effect for complete rotation into the hard direction is less
than 5 X 105 ergs/cm3 for Fe-Ni alloys.
The free-energy change of the
austenite-to-martensite transformation is of the order of 109 to 1010
ergs/cm 3 which completely overshadows the magnetocrystalline effect.
3. The magnetoelastic energy arises from the interaction
between the magnetization and the mechanical strain of the lattice.
Kittel and Galt (8)have shown that this energy can be thought of as
the change in magnetocrystalline energy with strain. A crystal will
deform spontaneously in order to lower the anisotropy energy and we
observe this effect as magnetostriction. The strains involved are
10-6 to 10- 4 , and they exist with or without the applied field. The
field, however, can alter the value of the magnetostriction by changing
the magnitude and orientation of the magnetization.
The lowest energy state of the crystal occurs when these strains
are allowed to take place. When the domains are confined by other
domains or other crystals, the energy of the crystal is not allowed
to decrease and a certain amount of elastic energy is stored at the
boundaries. The normal Gibbs free energy is measured for a crystal
that has a domain structure in which the net magnetization is near
zero.
This means that it already contains many domains of closure and,
therefore, a certain amount of magnetoelastic energy. When a field is
applied to the polycrystalline specimen, the domains of closure disappear in favor of domains with dipoles aligned with the field.
This
eliminates the magnetoelastic energy of the domains of closure, but
we now have different grain-boundary constraints which could increase the
magnetoelastic energy.
Fortunately, the strains are small and the energies
involved are negligible, i.e. of the order of 10 ergs/cm 3 in iron.
4. The magnetostatic energy, the self energy, the demagnetization energy, and the free-pole energy are all one and the same
thing.
In one line, it is the energy of the magnetic field generated
by the dipoles of the specimen, and the interaction of its magnetization
with that field.
Even though
an isolated magnetic pole does not
exist, mathematically and conceptually, it is sometimes convenient to
consider the magnetization of the material in terms of free poles of magnetism at opposite ends of the specimen.
These surface poles produce
a field within the material of value, Hd , which is antiparallel to the
magnetization. The energy of interaction of the magnetization with
this demagnetizing field, - 1 A+d'ddV, is only readily calculable for
one general shape, the ellipsoid. Other more complex shapes are very
difficult to treat because of nonuniform demagnetizing fields and internal free poles.
This leads to serious complications, such as nonuniform
magnetization within the specimen.
(It is necessary to assume uniform
magnetization in order to derive equation (4).)
For the ellipsoid,
Hd= - NM where N is the demagnetizing factor varying from zero (along
needles) to
4u
(normal to discs).
The magnetostatic energy per
unit volume is then:
Ems = 1/2NM2
(9)
This energy is sizeable and cannot be ignored until the applied field,
Ha , is much larger than 1/2NM (i.e. 1/2Hd).
gauss cm3 and N = 4w, we see 1/2NM = 20 kOe.
Taking Ms = 1700 ergs/
Therefore, at lowest
applied field of 20 kOe in the present research, the demagnetization
effect can be
the same order of magnitude as the magnetic contribution
to the Gibbs free energy. N, however, is usually much lower than the
maximum value of 47.
Since the bulk austenite in the alloys under study here is
paramagnetic or weakly ferromagnetic, its value of the magnetization is
small enough to make Hd insignificant compared to the level of our
applied fields.
In addition, N is much smaller than the maximum value
of 4w, inasmuch as the adopted specimen is rod-shaped. Although,
there are free poles at grain boundaries, these are usually considerably fewer in number than the free surface poles, especially when the
domains are aligned with the external field. A ferromagnetic austenite
spontaneously lowers its energy by creating domains of closure and
by aligning the domains from grain to grain.
the domains
to the field.
When the field is applied,
of closure are consumed by domains oriented nearly parallel
This causes an increase in the magnetostatic energy at
grain boundaries and other surfaces.
The exact energy changes are too
complex to calculate. Fortunately, this energy is small compared to the
Gibbs magnetic-energy change of the transformation.
The magnetostatic energy of the strongly ferromagnetic martensite
phase is also complex. If we approximate the martensite morphology
by oblate spheroids with large radius-to-semithickness, r/c, ratios,
and if they are magnetized parallel to their major axis, r, the
demagnetization factor, Nr , approaches zero.
If,however, they are
magnetized parallel to the semi-thickness axis, c, Nc can approach
4 r.
For an oblate spheroid, the energy difference between a plate oriented
parallel with, versus perpendicular to, the field is (11)
21
(10)
AEms = 1/2(AMs)2(N c - Nr )
Substituting Nc = 0.926, Nr = 0.037 (for r/c = 20) (12)
and AM =
1400 erg/gauss cm3 , (the difference in magnetization between austenite
and martensite), yields a value of 1.1 X 10 ergs/cm 3 for AEms.
This
energy is two or three orders of magnitude smaller than the freeenergy change of the transformation; however, it is no less than 10%
of the contribution of the magnetic Gibbs term for applied fields of
20 to 140 kOe.
The martensite morphology should, therefore, have the following
tendencies:
r will tend to align parallel to H and the r/c ratio of
plates parallel to the field should be greater than that perpendicular
to the field.
However, as the volume fraction of the martensite
increases, the free poles of the individual plates interact and the
orientation dependence should become less significant.
5. The Block or domain-wall energy arises from the exchange
energy and magnetostatic effects at the boundary.
When a high field
is applied to a domained structure, the favorably oriented growing
domains eventually eliminate the domain walls.
nal energy of the material.
This lowers the inter-
Block-wall energies are of the order of
1 to 3 ergs/cm 2 and are negligible for present purposes.
In iron, the
wall is about 120 atoms thick and has an energy of 2.9 ergs/cm 2 . This
16
3(7)
5
3
is equivalent to an energy density of n 106 ergs/cm3 () ( 10 ergs/cm 3
for nickel), and could be a factor during nucleation.
However, small
particles (, <200A in diameter) are probably composed of single domains;
therefore, the domain-wall energy does not enter the nucleation calculations, except perhaps as a small contribution to the surface energy
of the martensitic interface.
We have now shown that equation (7)is adequate to describe
the free energy of each phase.
The total free-energy change per unit
volume of the martensitic transformation is then given by:
AgT = Ago
where Ago
-
.oi••A
(11)
is the usual Gibbs free-energy change and A~ =
a
the difference between the magnetization of the two phases.
dynamically,
- iy i.e.
Thermo-
the other energy contributions, except possibly the
magnetostatic energy, are negligible with respect to AgT.
summarized in Table 1.)
(This is
On the other hand, we cannot say that the
kinetics of the reaction are not altered by these complications.
TABLE 1
Relative Magnitudes of Various Magnetic Energies
Type of Energy
Energy Range, ergs/cm 3
- 1010
Agoa'-y
10
Ag magnetic
107
magnetostatic
(shape anisotropy)
0
- 107
magnetocrystalline
0
- 105
magnetoelastic
0
- 104
domain wall
105 - 106
108
(1 - 3 ergs/cm 2)
_ _
1-3.
__
___
__ __
__
_ _
Advantages of Magnetic Fields for Research on Martensitic Transformations
Magnet fields, like hydrostatic pressure, offer another thermodynamic
variable to study martensitic transformations.
For a strong influence by
a magnetic field, there must be a large difference in magnetization between
the parent and product phases.
Martensitic transformation in steels more
than satisfy that criterion.
Pressure can be used to suppress the martensitic transformation because the molar volume of the martensite is greater than that of the austenite.
Although the experimentally attainable effect of pressure is
greater than that of magnetic fields (1 kbar = 21 kOe for y to a at
300 0 K in pure iron), obscuring side-effects can enter when pressure is
applied to influence a reaction that takes place by a shear mechanism.
It is also experimentally easier to measure the physical properties of
a crystal in an applied field than one confined in a pressure chanber.
In order to calculate the free energy of a phase at high pressures,
it is necessary to know the molar volume of both phases as a function
of pressure.
This is difficult to measure, and it is usually assumed
that the compressibilities of both phases remain unchanged under pressure.
To interpret magnetic experiments, we need to know the magnetization
as a function of field strength. This is readily measurable; in fact,
we have already shown that, for most cases, the saturation magnetization leads to accurate values of the Gibbs magnetic energy.
Magnet
fields, therefore, offer a unique opportunity to study the kinetics of
a thermally activated shear transformation without changing the temperature, pressure, or stress state of the material.
II. DISCUSSION OF LITERATURE
Martensitic Transformation Characteristics
II-1
The characteristics of the martensitic transformation have been
repeatedly documented in the literature. (e .g.13-14) The following discussion is merely designed to introduce the terminology, to bring up
the pertinent equations, and to pinpoint the critical issues.
II-1.1
Kinetics
A martensitic transformation is best defined as a diffusionless
(15-17)
transformation that exhibits a macroscopic shape-change. Attempts
have been made to further classify the reaction according to the observed kinetic behavior, i.e. how the reaction proceeds with time and
temperature. Much of the disagreement between these schemes has just
been a matter of semantics; however, there are some fundamental
differences and misconceptions.
Raghavan and Entwisle (R-E)(15)separate
the kinetic classification into three categories: athermal (defined as
transformations in which the progress of the reaction depends mainly on
falling temperatures), burst, and completely isothermal.
Magee and Pax-
ton (M-P)(16)define athermal transformations as those in which time
at temperature is not important, i.e. the total fraction transformed
should not be a function of prior thermal history. Stabilization effects
which occur in the "R-E athermal" alloys eliminate these from the M-P
definition of athermal.
They view the kinetics of ferrous martensitic
transformations as basically isothermal with the reaction rate having a
maximum with respect to temperature.
The difference in observed
kinetics are interpreted as stemming from the "parasitic" influences of
stabilization and autocatalysis (e.g. bursting phenomena).
Raghavan and Cohen (17) used the term anisothermal (18 ) to describe
thermally activated transformation during cooling to the test temperature
in which the activation energies for nucleation are so low that the
initial transformation cannot be measured or suppressed
(due to the
limitations of our measuring equipment and the experimentally attainable
quench rates).
They also define the athermal mode as transformation
without thermal activation corresponding to the situation in which the
activation energy for nucleation, AWa, goes to zero or to the level
of thermal energy available at the test temperature.
Although the R-E definition of athermal is convenient for practical
purposes, the question of whether all martensitic transformations are
thermally-activated isothermal reactions or whether there exists
true athermal modes at higher driving forces is fundamental to the
transformation theories.
From the Kaufman-Cohen-Raghavan model (19-20)
one would expect a true athermal behavior at temperatures below that
which
AWa approaches kT. Magee and Paxton (16) performed a critical
experiment to determine whether a distinguishable athermal nucleation
mode is operative.
They found that a temperature exists (= 135 0 K in
Fe 31.5Ni) below which the transformation rate decreases and there is
no evidence for athermal nucleation.
They, therefore, concluded that
the alloy exhibits thermally activated isothermal C-curve kinetics and
it is the elastic coupling between plates, i.e. bursting, that is largely
responsible for the accentuation of the transformation during cooling.
However, these experiments were not quite that clear-cut because the
investigators did not find a C-curve by straight quench-and-hold experiments, but by first quenching to 770 K and then upquenching to the
test temperature, using the method of Machlin and Cohen. (21 ) (They also went through a 770 K
+
room temperature
+
testing temperature cycle
in order to avoid the complication of anisothermal martensite formed
on upquenching.)
Magee ( 2 2 ) , in his small particle experiments, was able to observe
isothermal kinetics in a normally bursting Fe 22 Ni 0.49C alloy over
the temperature range of 163' to 223 0 K. Unfortunately, he did not
perform the critical experiment of transforming these particles at 770K
in order to look for the C-curve kinetics.
He would have answered these
questions once and for all.
The actual quantitative kinetic analysis has recently been reviewed
in detail by Ragahavan and Cohen. (17) nt,the number of most potent embryos existing at any time per unit volume of alloy, can be expressed as:
nt = (n i + p f - Nv) ( 1 - f)
= (n i + f [ p
where
( 1 - f
n.
= number of preexisting nucleation sites or embryos in
the parent austenite.
p
= number of autocatalytic embryos produced per unit
volume of martensite.
f
= volume fraction of martensite formed.
Nv
= number of martensite plates per unit volume of alloy.
= mean volume per martensitic plate.
The (1 - f) factor on the right hand side of Eq. 12 takes into account
the potential nuclei that are swept up by the transformed martensite
before they can nucleate.
The pf autocatalytic term (15) assumes that
the number of autocatalytic embryos created are proportional to the
volume fraction of transformed martensite, suggesting that elastic
and plastic strains set up in the austenite by the martensite might
be involved in the autocatalytic mechanism. pf works as well as or
better than other attempted (23) functional relations, using the fit
between the experimental transformation curves and those calculated from
these kinetic equations as the criterion.
The best up-to-date equation( 17 )
to describe the kinetics from measurements amenable to quantitative
metallography is the following:
df n
dt
ntvexp ( - AWa/RT) (V + Nv
-
)
(13)
v
where t is the time in seconds, v is the lattice vibrational frequency,
and AWa is the activation energy for nucleation at temperature T. This
equation contains many inherent assumptions, such as a single activation
energy for all nucleation sites and random nucleation events.
A detailed
critical analysis of Eq. 13 will be given in a later section of this
thesis.
If the mean plate volume is not a function of the amount of transformation, then -- equals zero in Eq. 13.
The Fisher partitioning
v
(24)
formula (24) predicts that as f increases, the transformed plates divide the austenite grains into smaller and smaller untransformed pockets giving rise to smaller and smaller plates. Pati and Cohen(25)have
shown that this overestimates the number of martensite plates required
to give a certain fraction of transformation.
If martensite were to
nucleate uniformly throughout the specimen, as Eq. 13 implicitly
assumes, then one might expect a Fisher type of partitioning to take
place.
In fact, however, plate clustering is commonly observed during
the early stages of transformation. The average plate volume within a
cluster might not be a strong function of f, if the cluster is simultaneously spreading into new austenite grains.
Obviously, the average
plate size must eventually decrease during the final stages of transformation when the clustering is complete and new plates are forced to
nucleate within the small pockets of retained austenite.
Ragahavan(26)did just such an analysis.
Following the Johnson-Mehl-
Avrami treatment and assuming that the autocatalytic effect is of the
same magnitude on both sides of the grain boundary, he considered the
transformation as a two-stage process: (i)the increase with time of the
number of austenite grains in which first-plate nucleation takes place,
and (ii)the progress of further transformation within such grains.
Using
this treatment, he was able to extend the fit between the calculated and
experimental transformation curves to higher percentages of martensite.
To be able to fit the whole transformation curve for a variety of testing
conditions with a minimal number of floating parameters-still remains as
the unfulfilled goal of the martensite kineticists.
Experimentally, McMurtrie and Magee(27)found that the average
volume of martensite plates was constant (= ~7x10
9 cm3
for a grain size
of 0.02 cm in an Fe24Ni 0.4C alloy) over a range of volume fractions from
0.07 to 0.55.
By assuming that the number of new plates per unit volume
of austenite per degree temperature change equals
where
c
,d(AGT 0
dT
.)
is a proportionality constant and d( Go
0dT') is the entropy
change on transformation, Magee( 28) derived:
) (Ms
T)
(14)
dT
q
where T is the quench temperature below Ms. A plot of In (1 - f) versus
ln (1 - f) = f
(AGv
(Ms - Tq) was linear up to (1 - f) = 0.05, implying V is effectively constant over this range if Eq. 14 is valid.
Contrary to the other workers,
Pati and Cohen(25)found that, in an Fe24Ni3Mn isothermally-transformed
alloy, the initial plate volume (f ~ 0.01) was largest near the nose
of the C - curve, and varied from 1 to 4x10 -10 cm 3 over the 600C range
on either side of the nose.
At about 30% martensite, V at all the reac-
tion temperatures approached a constant value of 1x10-10.
The apparent
discrepancy with other investigators may be due to differences in alloy
composition, however, it is more likely attributable to the fact that
Pati measured plate volumes very early in the transformation before the
steady-state cluster-spreading mode set in. Unfortunately, in the sensitive temperature range near the nose, Pati has no V measurements between 1 and 33% martensite; therefore, any rapid drop off of V with f is
only speculation at this stage.
An additional point related to this
topic is that one might not expect much variation in "plate" volume if
the martensite morphology is lathlike and grows in packets, as described
in the next section.
_ _
II-1.2
_ _
Morphology
Ferrous martensites are often divided into two major types - lath
(or packet) and plate (or lenticular) - which vary with respect to alloy
composition, temperature range of formation, crystallography, and fine
structure.
Lath (or packet) martensite, as the name implied, grows as thin
narrow strips, usually with the long direction parallel to with 60
of
[110]
related.
(29-30)
Adjacent laths having parallel widths may be twin
In these cases, the common interface is a (112)Y plane
which is closely parallel to the [ll0]y but ~ 5 1/2 degrees from the
long axis of the lath. The laths are typically aligned along a {ll1}
Y
plane with their widths and lengths approximately parallel to each other.
These planes of laths, in turn, are stacked parallel to one another,
forming what is usually referred to as the packet structure when observed
by optical microscopy. The habit plane of these packet planes is {111)},
but this may be considered as a psuedo-habit since the laths themselves
are nearly parallel to a {1121},
or {hhl}
type.
Lath martensite forms by slip and contains a high density of dislocations. Speich(31)estimated the dislocation density inside the laths to be
between 0.3 and 0.9x10
12
cm/cm 3 from electrical resistivity measurements.
Because of the close proximity of the planes of laths in a packet, the
austenite can transform to 100% martensite.
In contrast to the plate
(or lenticular) martensites, which can grow at one-third the speed of
sound in the metal, there is evidence(32 -33)
more slowly at speeds that can be filmed.
that lath martensite grow
31
Plate martensite, which is often idealized as double-convex lens
in shape, can have {3,10,15} ,
{259}y or
{225}y
habit planes. The
considerable scatter around these habits has been attributed by Bell
and Bryans (34 ) (for the {3,10,15}
habit) to the effect of prior trans-
formation and prestrain causing accommodation distortion during growth.
However, this is a matter of considerable controversy. The {3,10,15}type
of plates are well characterized by the phenomenological crystallographic theories, but the {225} type requires either a dilatation in the
habit plane or a deformation of the parent austenite(35)to fit the
theories.
The plates themselves often contain a midrib composed of a
regular array of {0l2}
0
transformation twins spaced 60 to 100A apart.
Shearing action appears to start on the plane of the midrib and is propagated in parallel but opposite directions on both sides of this plane.
Bokros and Parker(32)have demonstrated that the bursting phenomenon
commonly associated with the {259}y habit can be related to mechanical
coupling between plates, implying that autocatalysis is strongly influenced by the surrounding stress and plastic strain fields caused by
the existing plates. These plates are commonly observed to have an inner
(37-39 )
twinned region and an outer slipped region. Various investigators
have suggested that the change in the lattice-invariant mode from
twinning to slip is associated with a local temperature rise due to the
enthalpy of transformation.
Recent observations by Krauss and Marder (33 )
in which {259} martensite was tempered at 1570C and then requenched
demonstrated that certain plates widen on the second quench by a simple
extension of the outer slipped region with no apparent twinning. Since
the enthalpy heat of the first formed section had long been dissipated,
32
this shows that temperature is not the overriding factor, otherwise the
new sections would have formed by twinning.
Huizing and Klostermann (40 )
have pointed out that surface martensite, lath martensite, and the
untwinned part of a martensitic plate formed after a burst are all
morphologically different, but may be formed by a common mechanism.
Several alternate explanations have been offered to explain the
transition from lath to plate martensite. Owen, Wilson, and Bell( 41 )
suggested that if Ms is greater than T , the Zener ordering temperature,
lath martensite should form. However, low-carbon alloys having T. near
00 K should always form in the packet mode, but Fe 31 to 33 Ni alloys
transform to plates.(42)
The proposal (e.g.43)that low stacking-fault
energy favors the formation of lath martensite has also been rejected (33)
because both Ni and Mn favor plate formation, whereas Ni raises the
stacking-fault energy and Mn lowers it. The situation is not clear-cut,
however, in as much as the Fe Ni Mn alloys have a {225} habit which is
not well understood.
Increasing carbon or nickel favors the formation
of plates, but both of these alloying elements also lower the Ms
temperature. The temperature of transformation has been shown(e.g.33)
to offer the best correlation for the lath-to-plate transition.
Lower
temperature favors a twinning mode of transformation which presumably
favors plate martensite.
Recently, however, Davies and Magee( 42)were
able to produce plate morphologies at temperatures as high as 2100C
in low-carbon Fe Ni Co alloys. They demonstrated that packet martensite
always formed from paramagnetic austenite and that austenite ferromagnetism is a necessary but not sufficient condition for the formation of
plate morphologies in invar-type alloys. This effect was attributed to
33
increased flow resistance of the austenite due to ferromagnetic strength..
ening.
The strength or plastic behavior of the martensite and austenite
is emerging as a viable explanation for the transition inmodes of
transformation and is gathering a growing body of support. Related to
this line of reasoning, some investigators (3344)suggest that an
important criterion is the relative magnitudes of the critical resolved
shear stress for slip and twinning at a given temperature and composition.
On the other hand, Owen, Schoen, and Srinivasan(45)feel that the different
morphologies are the result of differences in the growth rate of the two
martensitic types, whereas Christian(46) feels that the growth mechanism
is determined by the nature of the interface. Both views can be selfconsistent if the growth rate determines the nature of the interface
or vice versa.
An additional idea by Bell and Owen(47)is that it is
necessary to exceed a critical driving force ( ~ 315 cal per mole) to
change from dislocated to twinned martensite. Pascover and Radcliffe(48)
applied this concept to other FeNi and FeCr alloys and found agreement,
except for Fe5Cr which should have been twinned but was not. Olson,(49)
inconsidering strain-induced nucleation, pointed out that we might
expect the lath morphology when the ratio of the nucleation-to-growth
rate is high and plates when the ratio is low. This suggestion could have
some merit, but itmay be that the actual event of changing modes in
the lattice-invariant strain plays an important role in the generation of
new autocatalytic nucleation sites.
34
11.1.3 Nucleation
At present, there are two schools of thought concerning the
crucial mechanism involved in the nucleation of martensite: The older
Kaufman-Cohen(19)model views the creation of new interface dislocations as the rate-controlling step, while other investigators, e.g.
Magee(28), believe that the motion of the austenite-martensite interface is rate-determining.
At this time, the Kaufman-Cohen-Raghavan(2 0 950) (KCR) model
(which is an improvement and an extension of the Kaufman-Cohen model)
is the only quantitative treatment of martensite nucleation, and for
this reason it is the only model used for quantitative comparisons in
this thesis.
In order to have a starting point, they assumed that
embryos greater than that needed for classical nucleation preexist.
They envision that the embryos could have formed at high temperatures
(at which the austenite is actually the stable phase) by the incorporation of existing dislocations. These embryos are then frozen-in
on cooling at temperatures near 300 0C and are ready to trigger-off
when brought into the Ms range.
This is very speculative, but just how the embryos originate is not
critical to the model. They then assume that the embryo is in the form
of a Knapp-Dehlinger(51)mini-plate having a (225)y Frank(52) interface.
Christian (53) has criticized the use of Frank's model of the interface
because of the experimental evidence that the close-packed directions
of the two phases are not exactly parallel and do not lie in the habit
plane. These observations are contrary to Frank's assumptions.
35
Christian (54) also pointed out that the Frank interface is essentially two
dimensional and predicts a large dilatational strain which is energetically
unfavorable and not observed experimentally in the fully grown plates.
Finally, Owen et al.(41)point
out that Frank's model of the interface
produces the lattice invariant deformation by slip, whereas electron microscopy studies have always revealed a finely twinned structure for
this habit, although not completely out to the final interface.
These criticisms seem damaging, and no one, including the authors, expects that the embryos are really well-formed particles of martensite.
Someone had to be bold enough to make this kind of assumption in order to
get the job underway. The KCR model is really a prototype model which todate has worked remarkably well to fit a wide range of data despite all
its shortcomings.
It has also evolved from a model of the initial
nucleation site to a kinetic model describing the growth of a plate. The
Raghavan-Cohen grow-path paper (50)expanded Magee's (16)analysis and considered the growth path of the embryo in terms of its radial-growth and
thickening kinetics. The predicted behavior of rapid radial-growth until
impingement followed by a slower thickening process is reminiscent of the
micrographs of plates showing a fast midribbed section and an outer
slower-growing slipped region. This is one of the encouraging results of
their model. A simple extension of the model to include twinning and heat
effects would be enlightening.
The critical step for nucleation in the above model is the creation
of a critical interfacial dislocation loop of radius p* at the tip of the
plate. This newly created loop is parallel to the already existing loops
which are perpendicular to the "flat" austenite-martensite interface. The
normal of the loop is in the [554] direction for a (225) habit. The
36
critical step of loop nucleation corresponds to growth in the radial
direction.
Plate thickening is envisioned to occur by the motion of
the screw components of the dislocation loop which are parallel to the
(225)Y habit plane and lie along the [110]y direction.
There is no driving force for thickening during the first few
radial-growth steps near the saddle point, but as the system moves
sufficiently along the free-energy surface, the free-energy change
per unit growth step in a given direction depend only on the partial
derivative of the free-energy change in that direction.
It is envision-
ed that the embryo is not at the saddle point but somewhat beyond in the
super-critical regime. The actual growth path is then determined by
kinetic factors, i.e. by the relative velocities of motion in the radial
and thickening directions.
These, in turn, will depend on the trans-
formational forces in these directions, as well as on the prevailing
kinetic barriers. The growth in the radial direction of the edge component of the existing loops is easier than thickening, yet it was assumed that the embryo maintains its oblate-spheroidal geometry during the
growth process. One might expect the Knapp-Dehlinger plate to be unstable and that the edge components of the loops would run off in the [ITO]
direction.
If this can happen, the Magee suggestion that dislocation
motion is the critical step would be substantiated.
The specific equations of the KCR model which were used in this
thesis to check the correlation with the data are the following:
AG =
AG
DA
-
2r cg
T
5 b2
-
2
4 rc2A + 27r2
rc'
*(6r) [nn
(p
(p*/b)
(15t
+ 0.4 + z]
(16)
37
5
i2
AU* = - 6 b2p* In(p*/b) - 1.6 + 2]
where
AG
(17)
= free-energy change attending the formation of a
martensitic particle oblate-spheroidal inshape
of radius r and semithickness c,
AgT = total (chemical plus magnetic)free energy per
unit volume given by Eq. 11,
Ac/r = strain energy per unit volume of the particle,
a
= interfacial energy per unit area which is considered coherent in the loop nucleation step but semicoherent for the bulk interface,
= shear modulus of the austenite,
b
= Burgers vector,
p
= critical-loop radius,
c'
= semithickness of the embryo at a distance 6r from
the tip,
6r
= incremental growth step of the plate in the r
direction,
=: core-energy parameter ~ 1.
Experimentally, the small-particle and the high-pressure experiments
have given us the greatest insight into the nature of the embryo. Electron microscopy observations are also valuable because they give us a
feel for the types of dislocation structures, stacking faults, and interfaces that are associated with the martensitic product. Much effort has
been directed to observing the embryo directly, but as Pati and Cohen(23)
pointed out, it is improbable that one exists in the small volume of an
electron-microscopy specimen under actual observation. Inone instance,
Venables (56) found a strain-induced embryo at the intersection of two
splates, and Olson and Cohen (57 ) used a double shear model to describe
the mechanism of its formation.
Detailed treatments such as these
throw a great deal of light on the nucleation mechanism.
The classical
small-particle experiments of Cech and Turnbull(58)
contain a number of simple yet revealing results.
By demonstrating that
some particles would not transform at the lowest cooling temperature,
they proved that the nucleation site for martensite has a heterogeneous
character, thereby eliminating all homogeneous nucleation theories as
viable explanations.
From the particle size at which about 1/2 of the
particles transform on quenching, we can obtain a rough estimate of the
numbers of initial embryos per unit volume ( 107cm-3).
their powders were single crystals.
About 3/4 of
By comparing the number of particles
containing grain boundaries in the transformed versus the untransformed
specimens, they were able to conclude that grain boundaries are not
important nucleation sites for the martensitic transformation.
Pati and
Cohen(23) came to this same conclusion using a much more indirect approach
on bulk specimens. Cech and Turnbull also reported that the particles
that transformed did so by bursting.
This is strong evidence that
autocatalytic embryos are newly created and not the triggering of prior
existing embryos.
Cech and Turnbull treated the martensitic transformation as an
athermal reaction, whereas Magee (22)studied the isothermal aspects using
polycrystalline atomized powders in the "as received" condition. As
mentioned previously, he was able to observe isothermal kinetics even in
his bursting alloy. He also showed that, when autocatalytic effects are
suppressed, there is no nucleation incubation time. Mechanistically,
this means that no detectably slow precursor steps occur in the
nucleation process.
In other words, at least some of the sites are
initially capable of immediately nucleating a martensite plate.
The concept of a single activation energy for initial embryos did
not fit the Magee (22 )data; rather the embryos were found to have a
distribution of effectiveness.
By assuming that the exponential law
for thermally activated processes is valid and that there is a constant attempt frequency, Magee calculated a distribution of activation
energies, AWa, to fit his data. The derived distribution was quite
unexpected in that it showed a negligible number of embryos with activation energies below a certain AWmin and then an equal number for each
additional energy increment thereafter. By using this AWa distribution,
Magee derived the following:
dNv
dv
= =AWa
f vexp( - Aa ) n*(AWa)dAWa
= n*(AWa)vRT
(18)
where n*(AWa)dAWa is the number of sites per unit volume having
activation energies between AWa and AWa + dAWa. Comparison with Eqs. 12
and 13 in Section II-1.1 indicates that ni = n*(AWa)vRT.
The experi-
mental values of n*(AWa)vRT range from 3x104 to 2xl05 cm-3 for FeNiMn,
whereas the most widely used previous estimate is 10 7 cm.-3 He also
noted that n*(AWa)v
increased with decreasing temperature and suggested
that larger driving forces could increase the density of effective
embryos. When Magee's tabulated data are plotted, however, (either n*(AWa)
or n*(AWa) RT versus T), it is apparent that even the magnitude of
the embryo-distribution function varies in a C-curve fashion with temperature. This may be intrinsic, but it would just as easily stem
from the assumptions made in the analysis.
The classical high-pressure experiments of Kaufman, et al.(58)
were interpreted as definite proof that martensitic embryos exist and
that they have a higher specific volume than the austenite. An Fe32.4Ni
austenite had been annealed at high pressures and then quenched under
pressure to room temperature.
The resulting Ms temperature at 1 atm
pressure was found to be greatly reduced, as the Kaufman-Cohen embryo
theory would predict. In addition, it was found that this effect could
be reversed, i.e. upon reheating above a certain freeze-in-temperature,
Kaufman, et al. were able to raise the Ms temperature of the alloy
back to nearly its normal value. These experiments seem to contradict
the results of Radcliffe and Schatz(59)who cooled Fe(O.3 - 1.2)C
austenites under pressure all the way from 9270C to the Ms temperature.
They concluded that the Ms depression was that expected from the change
in thermodynamic driving force with pressure, and that there was no
additional effect of a change in embryo potency. The discrepancies
between the two sets of experiments could be related to the large
differences in carbon content; however, interstitials should be mobile
at the higher temperatures of these experiments. Additional work is
needed to clarify this issue using gas-pressure rigs with low-carbon
bursting alloys.
11-2
Magnetically-Induced Martensitic Transformations
The application of a magnetic field to influence a metallurgical pro-
cess is not a new concept; it has already been used to alter texture(60-67)
order-disorder
( e .g.
68-69), precipitation(996070 -72), and spinodal trans-
formations. (73)
11-2.1
General Magnetic Transformation Characteristics
In 1929, E. Herbert (74) demonstrated that magnetizing a quenched
steel caused an increase in hardness.
This effect was not linked to the
martensitic transformation until 1960, when Sadovskii, et al.(75) reported
that they induced "intensive" martensite transformation in fine-grained
Fe 1.5 Cr 23 Ni 0.5 C at 770 K with a pulsed field of 350 kOe.
time, work in this area has flourished(7697).
Since that
The bulk of this effort
has been by Russian scientists using up to 500 kOe pulsed magnetic fields
on athermal chromium-nickel alloys.
(A pulsed field of 10-4 to 10-3
second duration is generated by discharging a capacitor bank through a
solenoid.
Most of the transformation occurs with the first pulse(77 ' 88)
the third and subsequent pulses give no noticeable change in the amount of
martensite.)
These results are summarized schematically in Fig. 2. Typi-
cally, the effect of the field is to raise MS in a linear fashion by some
0 C/kOe, as shown in Fig. 2a. At any given temperature, above
0.1 to 0.3
the normal MS , it is necessary to raise the field to a certain critical
value, HK , before any effect can be detected.
HK increases the MS tem-
perature to that of the testing temperature.
Fields greater than HK in-
duce greater and greater amounts of martensite, as depicted in Fig. 2b.
____~~
__
2000K
50
% a'
Ms
0
50
50
0
HK
300 kOe
H
200kOe
H
50
200kOe
HK
% a'
0
0
0
Ms
FIG. 2 SCHEMATIC
200 0 K
SURVEY
0
200 0 K
OF PAST PULSED FIELD WORK
The solid line is measured from a series of different specimens; however,
if larger and larger pulses are applied to the same specimen, we see a
stepped behavior in which the first jump at HK is the largest. As might
be expected, Fig. 2c indicates that higher and higher values of HK are
necessary to initiate an observable field effect as the temperature is
raised.
Finally, Fig. 2d illustrates the overall athermal transformation
characteristics for applied fields from 0 to 300 kOe.
mation curve is shifted to higher temperatures.
The whole transfor-
Turning on a field during
the course of a transformation effectively shifts the system to one of the
higher transformation curves.
Early workers ( 75 ) in studying the effect of H on the a' transformation
varied the grain size of their alloy in such a way that the finest grained
specimen had an MS just below liquid nitrogen temperatures.
When the field
was applied, they saw the largest increase in the amount of a' with the
finest grained specimen, i.e. in which there was no significant prior
transformation. Unfortunately, the interpretation of this effect was that
"the martensite had been 'artificially stabilized' by the small grain size
giving a sort of 'supermetastable' state which undergoes supercooling below
the normal position of the transformation point."
This reasoning led re-
searchers astray for a number of years. They were interpreting the pulse
as "destabilizing"' (81) the austenite or as "removing obstacles to the development of the potentially possible transformation under magnetostriction
stresses." (79) On the other hand, this viewpoint led to the investigation
of other forms of "austenite stabilization", i.e. that due to deformation
and prior transformation.
The only study of the effect of elastic deformation was by Fokina and
Zavadskiy(77) who showed that the combined effect of elastic stresses and
magnetic field were additive.
Plastic deformation of the austenite was
considered to suppress the a' transformation; thereby, requiring larger
values of HK in order to initiate the reaction.
If we have prior a', it
is usually much more difficult for the field (or any other driving force)
to increase the percent martensite; therefore, fields are not very effective in eliminating retained austenite.
Malinen and Sadovskii(8 5 ) have shown that the AS temperature for the
reverse (a'to y) transformation in Fe (26-31) Ni alloys is raised by a
steady field of 22 kOe. This is consistent with the increased stability
of the ferromagnetic phase in high fields (as long as AS < e C, the Curie
temperature).
II-2.2
Pulsed Versus Steady Fields
The application of a pulsed rather than a steady field could introduce
complications.
In fact, early investigators believed that the induced
stresses were triggering the reaction.
Since then, much Russian effort has
been devoted to proving that pulsed fields give the same results as steady
fields and that it is only the Gibbs magnetic term that is important. As
a rough approximation, they are correct.
Fokina et al.(83) showed that
an Fe 23 Ni 0.5 C steel in liquid helium transformed from 8-9% martensite
to 20-21% in either a 40 kOe pulsed or steady field.
In addition, predic-
tions based on the thermodynamic effect of field alone give reasonable
agreement with the observations.
The troublesome complications are all related to the pulse-induced
eddy currents.
Faraday's law predicts that an emf proportional to
be generated, opposing the applied field.
will
Pulsed fields, unfortunately,
have high values of the rate-of-change of field with time. Evidence for
the existence of eddy currents inthese experiments has been given by
Voronchikhin and Fakidov(84); the temperature of their specimens increased
with the square of the applied field.
gave a 50C temperature increase.)
(A 200 kOe pulsed field of 5 kHz
Eddy-current power losses are typically
proportional to B2 F2 where F is the frequency of the pulse and B is the
magnetic induction.
Aside from the heating effects, eddy currents lead to two other more
serious complications. The first of these has been called the "ponderomotive force".
In short(80), the induced eddy currents interact with the
surrounding magnetic fields leading to a stress proportional to H2 . On
pulsing in a solenoid, a cylindrical specimen is known to be subjected to
compressive forces normal to its cylindrical surface and tensile forces
along the axis
(
.
In strong fields, these forces can be quite high; in
fact, it has already been demonstrated( 98-99) that copper can be plastically deformed with a 350 kOe pulse.
Although no plastic deformation of
the higher strength steels has been detected, the stresses are still there.
Investigators were not able to alter the y to E (both paramagnetic) martensitic transformation in Fe(14-21)M (87-94) or the y' to 8 thermoelastic
martensite in Cu 14 Al 4 Ni(89 ) by pulsing with high fields.
They con-
cluded that the overriding effect is not the induced stress, but rather
the magnetic contribution to the Gibbs free energy.
Another complication of the eddy currents, known as the skin effect,
stems from the opposition of the applied field by the eddy current field.
The net effect is that the internal field varies within the specimen, being
maximum at the surface (= Happlied + 4rM - Hd) and minimum in the interior
( = Happlie d + 4rM - Hd - Heddy).
This, in turn, introduces second-order
complications, such as nonuniform magnetization and demagnetizing fields.
Malinen, et al.(87) plated the surface with a thin layer of high conductivity
copper. The enhanced eddy currents in this layer reduced the effec-
tiveness of the applied pulse for inducing martensite by a factor of two!
Finally, Fakidov,(80) et al. found that the critical field to induce martensite, HK , was a function of the applied pulse frequency.
The effect was
sizeable; HK varied from an extrapolated value of 50 kOe at F = 0 to 100 kOe
at F = 2 x 104 Hz.
The exact functional relationship, HK = H0 + 5F 0. 2 5 , is
critical because ponderomotive stresses should reduce HK with increasing F,
whereas if the skin effect were dominating it should reduce the effective
field and, therefore, HK should increase with F (as observed).
Thus, skin
effects have been shown to be more important than stress effects; however,
both exist and the relative contribution of each is a complex function of
the specimen shape, conductivity, elastic and plastic properties, and the
reaction being studied.
Even though pulsed fields are complicated by eddy-current effects,
they do offer some advantages over steady fields.
They are more economical
to generate and may lend themselves well to certain commercial applications.
Also, by using a single pulse, we can interrupt the martensite transformation at an early stage never before possible.
11-2.3
Isothermal Transformation Studies
There has been relatively little magnetic work on FeNiMn alloys which
will isothermally transform.(81-82,86) The first study by Estrin(82)
showed that the application of a steady field of 18.6 kOe did not cause a
sudden increase in the overall amount of martensite, rather there was an
increase in the rate of transformation.
He intuitively interpreted this
as a shift of the transformation C-curve to shorter times and to higher
temperatures, but he gave no evidence to back up this reasoning.
There has never been a study of an isothermal transformation by applying a large number of pulses and observing the transformation as a function of accumulative time.
Presumably, with a pulse length of 10- 4 seconds
and allowing time for the capacitors to charge up, the experiment would
take an inordinate length of time. This technique could still be a valuable tool for interrupting reactions with very high initial nucleation
rates.
Malinen, et al.(86) applied a pulsed field of 360 and 400 kOe to
Fe 23.6 Ni (3.3 and 3.6) Mn alloys which had normal isothermal nose temperatures of - 100 and - 140 0C respectively.
The field accelerated the
reaction, but the extent of transformation in both alloys rose monotonically with decreasing temperature (as depicted in Figs. 2b and d) with no
evidence of any kind of maximum at the nose temperature.
They unsuccess-
fully tried to apply nucleation kinetics to their results and finally interpreted the effect as a shifting of the transformation kinetics from the
isothermal to the athermal mode.
Magee and Paxton's
( 1 6)
This line of reasoning is contrary to
interpretation that all martensite transformations
are basically isothermal.
The most recent isothermal work has been by Peters, Bolton, and
Miodownik(96) on Fe 26 Ni 2 Mn with a steady field up to 20 kOe over a
temperature range of - 60 to - 800 C. They varied the fields during the
course of the reaction, and used the Kaufman-Cohen (19 ) model to calculate
values of surface energy and critical embryo radius as a function of temperature.
They did not dope the surface of their alloys to suppress the
surface complications, and therefore could not study the initial nucleation
rates.
They circumvented this difficulty by extrapolating a plot of log t
versus f, the volume fraction of martensite, to f = 0, and using the slope
and intercept of this plot to calculate the initial nucleation rate (assuming the average plate volume 9 to be constant).
Since autocatalysis from
the surface undoubtedly wiped out their initial transformation regions,
the log t vs. f plot looks somewhat linear over the small range they tested;
in contrast, if this plot is made for a doped specimen, it is sigmoidal in
nature.
(Replotting their data shows the expected curvature.)
Ignoring
these complications, the K-C model was found to fit the data well and the
major effect of the field was to vary the thermodynamic driving force.
II-2.4
Magnetic Effects on the Microstructure
Visual inspection of the microstructures shown in the literature (75'
87-88,92-93) does not reveal any directional alignment of plates or
changes
of morphology with the applied field.
(The e to a' transformation looks (94 )
somewhat directional, but no indication was given of the direction of the
applied field.)
Malinen et al.(87) noted that new plates appeared in pre-
viously untransformed austenite rather than around prior martensite.
This
was used to counter the arguments that induced magnetostresses around prior
plates might be significant.
Bernshteyn et al.(10) examined the isothermal y to a' transformation
while tempering Fe (5-16) Ni (0.03-1.1) C alloys under a steady field of
16 kOe. They reported (without micrographs) that in almost every case
there was an increase in the etchability of the boundaries of the martensitic crystals.
Undefined "statistical" treatment of the results revealed
that the crystals themselves were thinner, longer and showed less random
orientation (as we might expect from magnetostatic considerations discussed
in Section 1-2).
These structural differences were greater for the higher
nickel concentrations.
It is possible that at the higher temperatures of
this study, transformation stresses were less dominant; however, Fokina et
al.(13) transformed normal carbon steels in this same temperature range
with no observable orientation effects.
Since Bernshteyn had to use a
"statistical" analysis and since no micrographs were shown, the morphological changes are probably difficult to detect with the naked eye.
Anisotropy in transformed specimens has been indirectly detected by
Yermolayeo et al.(11) with magnetic torque measurements. The specimens
which were cooled without a field remained magnetically isotropic, while
those cooled in the magnetic field acquired uniaxial anisotropy. Assuming
that this resulted from magnetostatic-energy shape effects (as discussed
in Chapter I), they found that, as the total amount of martensite increased
from 4 to 20% (by using higher and higher fields), the total amount of
oriented martensite varied from 2 to 4%.
to about a factor of two.)
(These calculations are valid
These experiments show that anisotropies do
exist, and the best way to detect them is during small amounts of transformation before strain energies and other constraints begin to override.
It
is interesting to note that the authors were not able to detect these anisotropies metallographically.
II-2.5
Theoretical Treatment to Date
The martensitic transformation is not governed entirely by bulk thermodynamics as evidenced from the strong effects of specimen history; however,
most investigators have concentrated on the effect of magnetic fields on
the thermodynamics of the transformation and have neglected or misinterpreted the kinetic aspects.
II-2.5.1
Thermodynamic Predictions
Meyer and Taglang (10 0 ) were the first to derive a "Clausius-Clapeyron"
type of equation for magnetic fields:
AT0 -L
0
HAI
AS
TOHAI
L
(19)
where TO is the equilibrium temperature, AS is the entropy change on transformation, and L is the latent heat or the enthalpy difference between the
two phases.
(A more rigorous derivation of this equation is given in Ap-
pendix I.) This equation only applies for reversible equilibrium of phases
with heat capacities and magnetizations that are not strong functions of
temperature and field. Satyanarayan et al.
( 9 2)
derived the same equation
using similar triangles on a linear Ag versus temperature curve, assuming
that the other contributions to the thermodynamics such as the strain
energies, AgE , are not dependent on the field. This linear Ag (or constant heat capacity) assumption is crude since most free-energy curves show
HAl (TO-MS)
A
0
AgC
linear free-energy curves, but Ag
some curvature.
(AT
gives a better approximation(95 for nonis a hard number to pin down.)
If we
neglect these complications, Eq. 19 predicts that the greatest effect of
field will be on alloys with a large AI and TO and a small latent heat.
Unfortunately, TO/L does not seem to be a strong function of anything;
therefore, AI is still the most important material parameter.
Equation 19 does not strictly apply to the MS change of the irreversible martensitic transformation, yet it successfully predicts the linear
dependence of MS on the field depicted in Fig. 2a.
Investigators(82984992)
have become so confident in this equation that they have employed it to obtain quick and easy estimates of the enthalpy and entropy changes:
Voronchikhin and Fakidov(8 4 ) vacuum insulated a specimen and pulsed a field
through it while monitoring the temperature. After subtracting out the
eddy-current effects, this simple form of calorimetry (which does not involve the usual cooling technique) gave values of the latent heat (AH)
within 5 to 9% of those calculated from the MS temperature shift using
Eq. 19.
Eq. A-1 in Appendix I shows that the measured AS, the entropy
change of the transformation, approaches the correct value as the induced
AT
dT
temperature-shift becomes smaller; i.e. as 4-approaches dH.
To conclude, the magnetic Clausius-Clapeyron equation reflects the
behavior of the athermal martensitic reaction, but it must be used with
caution. Unfortunately, these thermodynamic equations have been repeatedly
invoked to interpret kinetic effects.
11-2.5.2
Kinetic Predictions
Malinen et al.(86) gave the first quantitative treatment ol
tion kinetics under a magnetic field while unsuccessfully trying
scribe the pulsed transformation of an isothermal alloy.
He ass
martensitic reaction was governed by an equation of the form:
N = k exp [- AW+U
WRT
where AW and U are the activation energies for nucleation and gr
spectively.
The ratio of the nucleation rate with the applied I
to that without, N, is (assuming U is small or, at least, insens
the field):
NH
-- =-exp
N
(AWH-AW)
[-
RT
where AWH is the activation energy for nucleation in the presenc
field.
Malinen used the classical value for the nucleation ener
which is inversely proportional
toAW (Ag)
NH
(1 4 - , yielding:
)44
- exp [
O
N
(AgO +POHM)
Peters, Bolton, and Miodownik followed the identical treatment a
but plugged in the Kaufman-Cohen value for AW = K1 + K2Ag in whi
K2 are functions of the surface energy, elastic constants, and the embryo
potency. This yielded:
NH
MH
= K
K2 [= ]
(23)
N
which accurately described their isothermal transformation kinetics under
a changing field.
II-2.6
Potential Applications
Magnet fields can be used to alter the physical and mechanical properties of a steel through the martensitic reaction.
Under most circum-
stances, magnetic fields will have no significant effect.
It is necessary
to be close enough to the MS temperature so that the value of the critical
field, HK , required to initiate the transformation is not too large.
In
many cases, if we were this close to MS , it would be cheaper to quench the
part to a lower temperature. Also, if we have greater than about 60%
martensite, further reduction in retained austenite becomes increasingly
difficult (both by increasing H or by lowering T).
Most experiments have
shown that the application of a field to reduce the retained austenite is
not worth the effort(e.g. 75); and, in particular, if the martensite is
generated by quenching to subzero temperatures, the application of a field
at room temperature appears to have little effect.
Magnet fields, however, do offer certain advantages.
It is much
better suited to mass production techniques than are low-temperature
quenching methods.
One can envision miles of continuous cast rod whipping
through the core of a high-field magnet. There has been recent interest
in magnetic-pulse fabrication of sheet-body parts for cars.
With the
right choice of alloys, one could simultaneously deform and transform.
Fields also have other advantages, such as minimizing quenching-stress
gradients and the ability to transform only a local section(75,88) of a
large part by applying a field only to that area.
If an FeNiC alloy(10289)
is quenched through a field to produce martensite and then tempered (or if
it is given a normal quench and then tempered in a field), the martensite
decomposition during the first stage tempering is suppressed.
If the tem-
pering temperature is raised to a point where the retained austenite decomposition begins, the field accelerates the reaction more than one would
predict from the thermodynamics.
Obviously, tempering in a magnetic field
is an area that still requires attention.
III.
PURPOSE OF THESIS AND OUTLINE OF WORK
The objectives of this thesis are to use magnetic fields as
a tool in order to investigate the nucleation and kinetics of the
martensitic transformation as well as to examine the intrinsic effects
of the field itself. The alloys chosen for this study were designed
to cover the range of morphologies from lath (packet) to plate
(lenticular) and the range of kinetics from isothermal C-curve behavior
to bursting. Particular emphasis is placed on the underlying causes
for these transitions in morphology and kinetics.
Since temperature
and magnetic field can be varied independently, the effects of chemical
driving force and temperature can then be studied separately.
Further-
more, using the magnet field as an additional thermodynamic variable
places a further test in the evaluation of current theories of martensitic nucleation.
The research itself is divided into two broad categories: (i)
Comparisons of the kinetics and morphology between zero-field and fieldaccelerated transformations, and (ii)High- and low-temperature magnetic
annealing and cycling experiments designed to influence the prior
martensitic nucleation sites.
IV. Materials and Procedure
IV-1
Specimen Preparation and Composition
The compositions and designations of all alloys used in this inves-
tigation are listed in Table 2. The first alloys were cast at the U.S.
Steel Laboratory for Fundamental Research in the form of 17-pound cylindrical ingots and hot rolled to one-inch slabs with intermediate anneals
at 1200 0 C.
(Microprobe analysis revealed a periodic variation of com-
position with a wavelength of about 150 microns.
Chemical fluctuations
were typically + 0.4% at 72% Fe, + 2%at 28% Ni, and + 0.6% at 3% Mn*.
On a relative basis, the manganese composition fluctuations were the
most severe). The second set of alloys was cast at the Ford Scientific
Laboratory in the form of 15 to 20-pound tapered cylindrical ingots.
These pieces were hot forged to 1/2-inch diameter rods at the AMMRC
Laboratory in Watertown introducing intermediate 900 0 C soaking treatments when necessary.
Rods 0.75-inch and 0.38-inch in diameter respectively were machined
from the above two sets of alloys.
These rods were swaged approximately
30% and then vacuum (l0- 5atm) encapsulated in quartz for a 3-day homogenization treatment at 1300 0 C. After this anneal, 0.1 inch was removed
from the diameter in order to circumvent any surface-composition complications.
The stock was then swaged down to 0.070-inch wire, with 30
minute 8000 C softening anneals after about every 70% reduction in area.
The resulting wire was cut into 1.2-inch long specimens and the ends
*All composition and alloy descriptions are given in weight percent, unless
otherwise indicated.
TABLE 2
Composition* of Alloys in Weight Percentages
Designation
Ni
C
S
USS 1439
28.7
0.0058
FORD V925
29.6
0.0090
0.005
FORD V927
30.8
0.0054
0.003
USS 1345
22.5
4.0
0.0055
0.00029
--
0.003
0;0006
0.0089
--
Balance iron
1
ground flat to a smoothness of 320 grit with a slight bevel on the edges.
At this point in the processing, the grain size was established
using a 1-hour 9000C anneal in evacuated vycor capsules.
Eight-inch
nickel (>100 ppmC) electrical resistivity leads were batch homogenized
in 0.2 atm CO for 2 hours at 10000C to insure reproducible carbon contents.
These leads were spot welded near the ends of the specimen
which was then electropolished for 2 minutes, with 30-second intervals,
in a 4% perchloric - 96% glacial acetic acid solution at 90 to 120 volts.
The surface was carbon doped by vacuum encapsulating the polished
specimens (together with the nickel lead wires), and annealing for 14
minutes at 8000 C. The assumed doping mechanism is the transport of
carbon from the nickel leads to the specimen via CO gas. (CO gas by itself (0.3 to 1.0 atm) was also found to be an effective dopant, but
was usually not used in this study.)
The specimens were air cooled after
this anneal and were usually subjected to transformation runs within
12 hours.
IV-2
Electrical Resistance Measurements
Relative resistance changes measured with a Kelvin double bridge
were adopted to follow the course of the isothermal reactions.
It was
necessary to measure [a] the initial (R25 I) and final (R25F) resistances
at 250C, [b] the initial untransformed (RI) and final (RF) resistances at
the testing temperature, and [c] the time-varying resistance (R)during
the course of the reaction at the test temperature.
If the incubation
time happened to be less than the first measured time for transformations
occurring at subzero temperatures under no field conditions, RI could still
be accurately estimated from the measured temperature coefficient of
resistivity. From these observations, the percent martensite as a
function of time was calculated from:
(R2 5F - R2 5I ) (RF - R)
A R25 1 (RF - R )
(24)
where A is the conversion factor equal to the AR25/R2 5 1 change
corresponding to a 1% transformation from austenite to martensite
(determined by comparing the 25%0
resistivity change with the amount of
martensite measured by pointcounting).
A was found to be 6.96 x 10- 3
7.78 x 10- 3 , and 3.10 x 10- 3 for Fe 28.8 Ni, Fe 30.8 Ni, and Fe 22.6
Ni 3.9 Mn respectively.
The Kelvin bridge was capable of detecting at 5 x 10-6 ohm change
or less. For a typical specimen resistance of 0.01 ohm, this corresponded to about 0.1% martensite. Unfortunately, a 1/20C temperature variation
could also cause this order of change in resistance.
small ( AR
In addition, a
_ + 2 x 10- 4 ) increase in resistance was observed due to
magnetoresistance effects of the field.
This does not complicate matters
since Eq. 24 shows that we were always dealing with resistivity differences
at the test temperature.
If,however, there were some transformation
during the interval before the first resistivity point was taken, it was
necessary to add the small magnetoresistance-increase to the resistance
measured at the test temperature just before the field was applied for an
accurate estimate of RI .
In order to avoid undue ambiguities, 0.3% martensite was used to
mark the "beginning of the reaction".
The reaction time was started
when the field reached about 95% of its final value and the first
resistivity point was taken within about 0.3 of a minute. This put
a practical lower limit on the measurable incubation times.
Any re-
ported times less than this are linear extrapolations from longer times.
IV-3
Temperature and Magnetic-Field Control
Initially, subzero temperatures were obtained using the standard
baths of water and ice, menthanol or acetone and dry ice, or 75%
petroleum ether - 25% methylcyclohexane and liquid nitrogen. Unfortunately, there was only about one inch of working space inside the dewar
in the magnet, and inadequate stirring led to temperature fluctuations
of up to + 20C. It was necessary to build a low-temperature controller
shown diagrammatically in Fig. 3 consisting of a copper pipe closed at
one end and wound with a bifilar brass-wire heater. In operation, the
insulated tube was filled with one of the previously mentioned solvents
and stirred with a stream of cooled helium or nitrogen gas.
The whole
assembly was immersed in a cooling bath of liquid nitrogen or dry ice methanol which was also used to pre-cool the stirring gas.
Two thermo-
couples were positioned within 1/8 inch of the specimen to monitor the
temperature and for input into a temperature controller. The temperature
controller itself was converted to a low-temperature controller by using
a liquid-nitrogen cold junction.
With this setup, a 200C subzero-tem-
perature change could be completely equilibrated within three minutes,
0 C over the temperature
and the temperature could be controlled to + 0.2
range of + 100 to -140 0C.
High temperatures (3000 to 9000C) were obtained by means of a spiral
)IL
FIG.3
SCHEMATIC OF CRYOGENIC TEMPERATURE
CONTROLLER USED DURING TRANSFORMATION
IN APPLIED FIELDS.
silicon carbide globar inserted from the bottom of the magnet and
surrounded by a water-cooled copper jacket. A 14mm, Vycor tube was
placed in the center of the globar, which served both as a support for
an encapsulated specimen and as a channel for an air-blast quench. The
temperature was maintained constant over a 2-inch distance to with +
5oC and the specimen could be air-blast cooled to 200 0 C in about 20
seconds.
The magnetic fields for this study were generated at the Francis
Bitter National Magnet Laboratory by passing up to 25 kamps through
special axial-flow helical solenoids.
The fields varied from 20 to
140 kOe for times up to 3.5 hours in a 2.1-inch bore magnet.
It took
approximately 15 seconds to turn up the field to 140 kOe because of
control-room restrictions, however, even if we could have turned the
field up faster, it would have taken about that long for eddy-current
effects to die down.
IV-4
Magnetization Measurements
The magnetization of the austenite and martensite of each alloy was
measured at the temperatures of interest as a function of field
strength, using a vibrating-sample magnetometer in a Janis Super-VariTemp dewar. The magnetometer was calibrated before each run with a
pure nickel sample having a known ( M = 54.4 erg/gauss gm) magnetization.
4.5 mm long sections cut from the center of the 0.070-inch diameter wire
specimens were vibrated parallel to the wire axis in a transversely
applied field up to 10 kOe.
For the martensitic samples, it was neces-
sary to subtract out the magnetization of the retained austenite assuming
the contributions from the two phases to be additive. The martensitic
phase comes very close to its saturation magnetization at 10 kOe; no
attempt was made to extrapolate these values to higher field strengths.
IV-5
Optical Microscopy Examination
The specimens to be studied were mounted longitudinally in Clear
Cast cold-setting compound. After mechanical polishing to Linde A
grit, they were immersed in methylene chloride for a few hours in order
to disintegrate the mounting material.
The specimens were then electro-
polished for 1.5 minutes, in 30 second intervals, in a 4% solution of
perchloric acid in glacial acetic acid at 90 to 120 volts and 0.6 amps/
cm2; the temperature range was 120 to 180 C.
The optimum polishing condition was obtained when a bubble-free
brown polishing layer was formed, which tended to "drip off" the surface.
The specimen was immediately rinsed in water and etched in a
solution of 250 grams/liter of sodium bisulphite in water.
This solution,
100 times more concentrated than usually used, brought out the structure
in 10 to 200 seconds depending on the alloy composition. (Intermediate
exposures to air during the course of the etch or an acetic acid dip
will accelerate the reaction.)
Careful control of the etching conditions
was found to result in orientation-dependent color contrast varying from
blue-reds to tan-browns.
IV-6 quantitative Metallography
IV-6.1
Grain-Size Measurements
Grain-size measurements were carried out on specimens heavily etched
to reveal the austenetic grain and twin boundaries. The mean linear
intercept, L, was determined by counting at least 800 intersections of
grain and twin boundaries while making four longitudinal passes along the
specimen length and recording the total distance traversed.
The grain size was converted to equivalent spherical diameter
using (101)
a = 1.65 L
a
(25)
for Fe 29.6Ni, Fe 30.8Ni, and Fe 22.6Ni3.9Mn were 0.0407, 0.0590, and
0.0220mm respectively. The FeNiMn ternary alloy had a factor of two smaller
grain size, which must be kept in mind when comparing its reaction
kinetics to that of the binary FeNi alloys.
IV-6.2
Volume-Fraction Measurements
The mean volume fraction of martensite, f, was determined by
superimposing a grid on a 500X or lO00X photomicrograph and counting the
number of point-intersections (out of a possible 1200) that landed on
martensite. (Boundary hits were counted as 1/2). This procedure was
repeated for 3 to 5 random photomicrographs and the results averaged.
f was used either to determine the conversion factor for the resistivity
change to percent martensite or as an input parameter in the average
volume-per-martensitic-plate calculations.
65
IV-6.3
Measurement of Mean Volume per Plate and Radius-to-Semithickness Ratio
The number, length, and maximum thickness of all plates within an
18 by 20 cm square area were measured on 3 to 5 random longitudinal
photomicrographs blown up to 2000X. This usually meant measuring from
300 to 600 plates for each specimen.
For specimens with larger volume
fractions of martensite, smaller representative areas (approximately
10 by 11 cm square) were picked out and all the plates within this
area were counted.
smallest plates.
Special care had to be taken to count even the
The measurements were recorded to the nearest
0.005 inch on a 2000X photomicrograph, which works out to be about 0.06
microns. Many times, the "plates" themselves were ill-defined, especially
in the Fe 29.6Ni alloy.
From these measured numbers and from the point-counted values of the
volume fraction of martensite, we computer-calculated E, F, (c), N
A,
Nv, and V, assuming the plate shape to be approximated by thin oblate
spheroids.
The definitions of these quantities and the derivation of the
the relevant equations are described in Appendix II.
Conservatively, the
absolute values of these calculated parameters are probably valid to
within a factor of 5; however, the relative values with respect to each
other should be good to + 10%.
IV-7
Measurement of Nucleation Rate
The initial nucleation rate was estimated from the relation proposed
by Shih et al:(102)
(26)
0.003 = NITi
where N is the initial nucleation rate, V is the mean plate volume, and
ri is the time to reach 0.3% martensite.
This equation is only an
approximation, since it incorrectly assumes that the rate of nucleation
and the average plate volume are constant during the incubation period.
The ramifications of these assumptions will be analyzed in a later chapter.
IV-8 Averaging Techniques
A very time-consuming problem encountered in this investigation
was the experimental scatter in the isothermal-transformation studies.
This seems characteristic of martensitic research of this nature. The
problem was resolved by repeating the experiments under identical conditions
from 3 to 15 times and picking the "most representative" curve which had
the median value of incubation time.
Rather than use this median incubation
time in Eq. 26 to calculate AWa, we felt it was more accurate to obtain
AWa for each full curve and then take the median AWa in this way.
V. RESULTS AND DISCUSSION
V-I
Material Characteristics
Before proceeding into the main body of the results, it may be help-
ful to examine the material characteristics of the alloys under investigation.
V-1.1
These are summarized in Table 3.
Shear Modulus and Density
The values of the shear modulus and density of these alloys in the
austenitic condition were measured by Speich at the U. S. Steel Fundamental
Laboratory. These data are presented in the form of computer-fit leastsquare equations as a function of temperature in 'C. The FeNiMn modulus
data were determined over the whole temperature range of interest down to
- 196 0 C; however, the binary FeNi moduli were not measured below - 200 C be-
cause of the ensuing martensitic transformation.
The actual values are
not as precise as shown in the table; these were the equations used to accurately reflect changes with temperature in our model-fitting exercises.
Note that the shear modulus of the FeNiMn alloy is greater than that
of the binary FeNi alloys.
The Mn alloy will, therefore, require more
energy to nucleate and grow since both dislocation core energies and strain
energies(51,103) are proportional to the shear modulus.
In the KCR model,
a large shear modulus can prevent the athermal-transformation mode from
ever occurring, thereby yielding isothermal C-curves (as observed in
FeNiMn alloys near the composition being studied in this thesis).
Also,
by making the assumption that the moduli of the austenite and martensite
I
1
--
_
-
TABLE 3
Material Characteristics of the Alloys Under Investigation
Grain Size
(mm)
Alloy
(wt. %)
~ Density (a'+y)
(gm/cm 3 )
AMY-ýa
Shear Modulus (y)
(dyne/cm2 )
Fe 28.7 Ni
(0C)
(ergs/gauss gm)
-10
20C
150.8 at + 90C
Fe 29.6 Ni
0.041
8.183
-3.2 x 10-4 T
6.840 x 1011
-1.11 x 108 T
-3.75 x 105 T2
30
148.1 at -
Fe 30.8 Ni
0.059
8.173
-3.2 x 10-4 T
6.678 x 1011
+1.53 x 108 T
73
123.3 at - 200C
<-196
162.3 at -196 0 C
165.3 at -120 0C
163.9 at - 810 C
-1.09 x 106 T2
Fe 22.5 Ni 4.0 Mn
0.022
8.148
-1.96 x 10-4 T
7.469 x 1011
-3.68 x 108 T
-7.75 x 105 T2
162.6 at - 200C
(42).
T is temperature
reference in oC for this table only.
From
From reference (42).
_
are additive in a duplex structure, Speich's data show that the martensite
modulus is at least 20% smaller than that of the austenite. This is in
(104) . These lower
agreement with the findings of Goldman and Robertson
values suggest that the martensite may be the phase to elastically and
plastically accommodate the strains.
This, in turn, could alter the kine-
tics and morphology of the transformation, e.g. self-accommodating strains
in the martensite could force the growing plate to change from the twinning to the slip mode.
The densities of the a' + y listed in Table 3 were determined from
transformed specimens containing from 50 to 90% martensite.
These least-
square fit equations were used as a rough approximation to the temperature
variation of density in the conversion of cal/mole to ergs/cm 3
V-1.2
Magnetic Properties
All three binary FeNi alloys are ferromagnetic at lower temperatures;
however, the Fe 28.7 Ni alloy can transform at temperatures above OCy in
the paramagnetic austenite region.
The FeNiMn alloy remains paramagnetic
down to liquid nitrogen temperatures in the austenitic condition.
Mossbauer studies on a similar alloy (Fe 24.2 Ni 2.98 Mn) at the U.S.S.
Laboratory revealed a broad split austenite peak at 120 K. This type of
pattern has been interpreted(105) (inan FeNiCu system) as an indication
of the occurrence of a mixture of antiferromagnetic and ferromagnetic
phases. This interpretation supports the two y-state ideas of Miodownik
(106), which were employed by Kaufman(1 0 7 ) in his calculation of the
chemical driving-force equations used in this thesis.
As discussed in Section 11-1.2, Davies and Magee 42) have correlated
(ec-MS ) with morphology.
We will show later that our zero field struc-
tures fit very well into their classification scheme; however, the fieldinduced morphologies reveal that their correlation does not given the
complete story.
The values of the magnetization at 10 kOe, shown in Table 3, have an
estimated accuracy of about 5% on an absolute scale.
In the Fe 29.6 Ni
alloy, AM is larger at + 90C than at - 20C because the former is just below
eCy; consequently, the magnetization of the austenite is a strong function
of temperature in this region. Since the Gibbs magnetic energy is proportional to HAM, the field is more effective at higher temperatures in this
alloy. The value of AM is lower in the Fe 30.8 Ni alloy because the
austenite is more strongly ferromagnetic.
The magnetization curves shown in Fig. 4 can yield additional information pertinent to the martensitic transformation.
The binary FeNi mag-
netization curves reach a plateau of saturation after a relatively sharp
transition region, whereas the ternary FeNiMn alloy saturates over a much
wider range of fields.
This behavior is reminiscent of the differences in
the shape of the kinetic curves in these alloys as they approach the final
stages of the martensitic transformation.
This comparison could be signi-
ficant since the motion of Block walls and the rotation of domains may
be somewhat similar to the motion of the martensite interfaces and the
atomic shuffles of the transformation.
It has been suggested (e .g. 33) that
deformation studies would be valuable in correlating the plastic properties
of a material with its martensitic-transformation behavior.
The variation
Wl
m
200
160
_120
E
80
40
00
0
2
H(kOe)
FIG. 4
8
6
4
ROOM TEMPERATURE
-
10
D
MAGNETIZATION
CURVES
of the shape and hysteresis of these magnetization curves as a function
of the rate of applied field could yield similar data but with a simple
nondestructive test. The FeNiMn magnetization curves have a smaller
hysteresis area than the binary alloys implying that they were more reversible in nature. Temperature has little effect on the hysteresis of
these curves; however, varying the sweep rate from 0.02 to 0.2 kOe/sec
increased the hysteresis by an order of magnitude for all three alloys
tested.
V-1.3
Chemical Driving Force
The chemical driving force, AG
O , (incal/mole) was calculated from
the chemical compositions, using Kaufman's(107) latest equations:
AG0 = (I-X-Y)GFe + XGNi + YGMn + X(1-X-Y)GFeNi + 1000 XY
(27)
+ (1-X-Y)Y(- 5950 + (6.17 x 10-3)T 2 )
where T is the absolute temperature in OK, X and Y are the atomic fractions of nickel and manganese respectively, and:
GFe = 1303 + (1.78 x 10-3)T 2 - (2.87 x 10-5)T 3 + 4.91 x 10-8)T 4
GNi = - 940 - (9.82 x 10-4)T 2 + (1.16 x 10- 6 )T3 - (3.37 x 10-10)T 4
GMn = 260 + (1.18 x 10-4)T 2 - (7.64 x 10-7)T 3
GFeNi = - 1350 + (1.48 x 10-4)T 2 + (2.02 x 10-7)T 3
73
The results of these calculations are plotted in Fig. 5.
The curves
flatten out at lower temperatures as expected from the third law of thermodynamics; consequently, lowering the temperature below 100 0 K does not give
much of an additional driving force.
It is also to be noted that the
driving force versus temperature behavior of the FeNiMn alloy is very similar to that of the Fe 29.6 Ni alloy, suggesting that the large differences in their transformation kinetics is not due to driving-force considerations.
-180
-220
- 260
o-
300
I
(9
<
-340
-380
-420
0
100
200
T (K
)
300
-
.
FIG. 5 CHEMICAL FREE-ENERGY CHANGE FOR THE AUST
TO-MARTENSITE TRANSFORMATION FROM EQUA'
V-2
V-2.1
Fe 28.7 Ni --
(Isothermal)
Kinetics
The three FeNi alloys investigated in this thesis cover the range of
kinetics from isothermal to burst transformations and the range of morphologies from lath to plate.
Fig. 6 shows the transformation behavior
of the Fe 28.7 Ni alloy under zero-field conditions.
(The earth's mag-
netic field is only about 0.4 gauss at the Magnet Laboratory.)
The reac-
tion becomes easily detectable at temperatures near 130C and, as the temperature is lowered, this alloy transforms at a continuously increasing
rate to higher and higher final percentages of martensite. Over the temperature range studied, isothermal kinetic behavior is unquestionably observed. As is characteristic for FeNi alloys, the transformation reaches
a relatively abrupt slow-down at the final stages.
The reasons for this
plateau behavior is not well understood.
Fig. 7 illustrates the accelerated kinetics under the influence of
magnetic fields at + 24.40 C, 120 C above the temperature of the first detectable transformation without a field.
Increasing field and decreasing
temperature behave similarly in that both raise the driving force and
accelerate the kinetics.
A closer comparison of the two figures reveals
that there are some differences.
In particular, the field-accelerated
curves reach their plateau faster than the zero-field curves at lower
temperatures.
This suggests that the plateau behavior is thermally acti-
vated and is enhanced at the higher temperatures of the field-induced
transformation. Later discussions will show that this approach is too
= f-%
Fe 28.7 Ni
NO APP LIED FIELD
t
uii
w
I-J
CHl)
bLJ
H(.)
uJ"
I0i
TIME MINUTES
"-
FIG. 6 TRANS FORMATION KINETICS OF Fe 28.7 Ni
UNDER ZERO-FIELD CONDITIONS
Fe 28.7 Ni
MAGNETIC FIELD APPLIED AT +4
50
40
it
Lu
30
~I-
20
bJ
H
H
z
10
(_)
bJ
w~
A
0
20
60
40
TIME MINUTES
8(
-
FIG. 7 FIELD-INDUCED TRANSFORMATION
OF Fe 28.7 Ni AT 24.4*C
100
simplified.
Could we have predicted the effect of magnetic field on the transformation curves from the thermodynamic driving forces and a knowledge of
the zero-field characteristics of Fig. 6?
It might seem reasonable to
assume that the kinetic reaction rate would be a simple function of the
thermodynamic driving force which we can increase by lowering the temperature or raising the field. At zero field, a transformation at + 130 C
has a chemical driving force Df 256.8 cal/mole.
Comparison of Figs. 6
and 7 shows that 40 kOe at 24.4 0C gives roughly comparable transformation
kinetics.
The chemical driving force at + 24.4 0 C is 244.5 cal/mole and
the magnetic p0MH contribution is about 9.1 cal/mole yielding a total
driving force of 253.6 cal/mole. Thus, we see that the field-induced
transformation occurs with a smaller driving force. At higher fields,
the difference is even more marked. Obviously, the thermodynamic driving
force is not the only parameter dictating the kinetics of the reaction.
V-2.2 Morphology
Iron-nickel alloys with nickel less than 28.5 wt. % have lath morphologies, while those containing greater than 30.5% exhibit plate characteristics. The alloys in the narrow band of composition in-between
exhibit what has been described(42) as mixed (not intermediate) morphologies. The morphology of the Fe 28.7 Ni alloy, shown in Fig. 8, is
just within this transition range. The structure is extremely hard to
characterize, to say the least.
It is definitely not plate-like, but it
79
p
50,u.
FIG.8
LATH-LIKE
MORPHOLOGY OF Fe28.7Ni
TRANSFORMED WITH A 60 kOe
FIELD AT 24.40 C.
80
does not exhibit the packet blocks of lath martensite, either. For future
designation, this will be referred to as lath-like martensite exhibiting
a loose-packet structure.
Because the transformation temperature was so
close to room temperature, surface artifacts due to mechanical polishing
and even surface-induced martensite during electropolishing hampered the
morphological investigation.
In addition, this alloy exhibited severe
clustering of the martensite in streaks parallel to the specimen axis.
Within these limitations, the field had no apparent effect on the morphology.
V-3
V-3.1
Fe 29.6 Ni - (Isothermal
+
Bursting)
Kinetics
The zero-field kinetics of the Fe 29.6 Ni alloy are shown in Fig.
9. The transformation can be observed within a reasonable length of
time at temperatures of about - 8 or - 90C. As the temperature is
lowered, the kinetics are accelerated, but remain definitely isothermal,
until a "transition" temperature near - 200C is reached at which the
transformation can be described as a very fast initial reaction followec
by a slower isothermal reaction.
One specimen at - 200C took about 30
seconds before it was possible to detect any transformation, and then it
abruptly transformed with a fast "burst" and audible clicking.
This
illustrates that even the fast bursting region is thermally activated.
At liquid-nitrogen temperatures, there is a much more dramatic burst
and a smaller "isothermal"
tail.
The effect of large magnetic fields on the transformation kinetics
at constant temperature is illustrated in Figs. 10 and 11.
Again, the
field-accelerated reaction yields a series of curves very similar to
those generated by lowering the temperature.
It is necessary to go to
higher fields at + 90C than at - 20 C because of the smaller chemical
driving force available at the higher temperature.
The final plateau
stage is reached at earlier times in the field-induced curves than in th
no-field curves generated by lowering the temperature. The effect, however, is not nearly as noticeable as in the Fe 28.7 Ni alloy tested at
+ 24.40C. The fact that this effect is less noticeable at + 90C than
Fe 29.6 Ni
NO APPLIED FIELD
100
LUJ
80
U)
60
LU
H
Z
0
ILU
I-
40
20
0
0
20 40 60 80 100 120 140 160 180
TIME MINUT ES )
FIG. 9 TRANSFORMATION KINETICS OF Fe 29.6 Ni
UNDER ZERO-FIELD CONDITIONS
Fe 29.6 Ni
MAGNETIC FIELD APPLIED AT -20C
100
LU
80
i-
60
Ea
0-LU"
z,0•
0uLd
40
20
0
0
FIG. 10
20
40 6 0 80 100 120 140 160 180
TIME (MINUTES )
FIELD-INDUCED TRANSFORMATION
OF Fe 29.6 Ni AT -2 0 C
KINETICS
Fe 29.6 Ni
MAGNETIC FIELD APPLIED AT+ 90 C
100
LU
80
H-
z
LJ
i-
60
H-
40
z
C)
LU
w:1
20
0
0
FIG. II
20
40 6 0 80 100 120 140 160 180
TIME (MINUTES)
FIELD-INDUCED TRANSFORMATION
OF Fe 29.6 Ni AT + 90 C
KINETICS
at +24.4°C might be taken as further evidence to support the hypothesis
that the plateau behavior is related to a thermally activated phenomenon;
however, discussions in Section V-3.3 will show that this is too simplified an approach.
V-3.2 Morphological Transition Behavior
Optical microscopy reveals some interesting correlations and some
new clues as to the intrinsic differences between lath and plate martnesites.
As mentioned previously, the Fe 29.6Ni alloy falls about midway
within the lath-to-plate transition composition range.
Under zero-
field conditions at temperatures from - 90C to - 150C, the structure
is lath-like, reminiscent of the Fe 28.7Ni structure.
As the temperature
is lowered, however, the structure changes to a mixture of plates (with
a definite midrib) surrounded by the higher temperature lath-like
structure, as in Fig. 12.
The morphological transition temperature of
the Fe 29.6Ni alloy is near - 200C, the same temperature at which we
saw the transition in the kinetic mode in Fig. 9. As the temperature is
lowered even further, the plate morphology begins to predominate.
This
is strong evidence to support the hypothesis (e.g. 16) that all martensitic
transformation kinetics are isothermal and the only difference between
bursting and isothermal alloys is the cooporative nature of the bursting
morphology. This also proves that the temperature of transformation is
more correlative with the lath-to-plate transition than alloy composition.
However, it should be emphasized that these relationships in Fe Ni alloys
do not necessarily apply to Fe Ni Mn alloys.
_ I~j___
(a)
F-
50//
(b)
FIG. 12
MORPHOLOGICAL TRANSITION IN Fe 29.6 Ni
UNDER ZERO-FIELD CONDITIONS
(a) LATH-LIKE AT -13 0 C
(b) MIXTURE OF PLATE AND LATH -LIKE
AT - 196 0 C
An examination of the morphology of the higher temperature fieldinduced transformations, shown in Fig. 13, reveals that they also go
through the same structural transition as the zero-field transformations.
This transition in the Fe 29.6 Ni alloy occurs roughly with a field of
50 kOe at
-2*C and with 90 kOe at
+90C. These findings demonstrate
clearly that temperature is not the only important factor in correlating
the lath-to-place transition. The
+90C/140 kOe specimens have fewer
plates than do the - 260C zero-field specimens with a comparable amount
of transformation.
However, this could be related to the fact that it
takes more time to raise the field to 140 kOe than to quench to - 260C.
The hypothesis that the total Gibbs free-energy change is a proper correlation parameter for the morphological transition seems appealing at first.
Comparison of Figs. 9 and 11 show that the zero-field/-9oC
and the 60 kOe/ +90C curves display nearly the same kinetic (and
morphological) behavior.
Their total (chemical plus magnetic) driving
forces are 257.2 and 249.7 cal/mole
respectively. Thus, just as in the
Fe 28.7 Ni case, the magnetically-induced transformations occur with
smaller driving forces.
So far, we have shown that the chemical composition, temperature,
and free-energy change do not correlate well with the kinetics or
morphology of the martensitic transformation. This situation should
come as no surprise, however, since it is rare that a complex kinetic
reaction can be described by a single thermodynamic parameter.
More
satisfactory correlations with kinetic parameters will be made in the
next section.
Because of the loose-packet nature of the lath-like martensite, it
88
I-~I
(a)
*"
50•
(b)
FIG. 13
FIELD-INDUCED MORPHOLOGICAL
TRANSITION
IN Fe 29.6Ni AT-2°C
(a)
LATH -LIKE AT 40 kOe
(b)
MIXTURE OF PLATE AND LATH-LIKE
AT 140 kOe
was practically impossible to analyze quantitatively, as has been done
for plate structures. (25)
Visual comparisons between the structures of
several specimens revealed that the lath-like structure was insensitive
to changes in temperature, field, percent martensite, and the simultaneous existence of plates. (The mean volume per plate was somewhat
larger when the martensite content was <5%).
Inaddition, although
austenitic twin boundaries usually behaved like grain boundaries and
stopped the growing martensite, there were examples of twin-related
loose-packets as shown in Fig. 14.
These transition structures have been referred to as a mixture of
plate and lath martensite, but they are more than that. We have
already seen that the lath martensite in the mixture is only lathlike; moreover, closer inspection discloses that the plates themselves
also have an intermediate structure. When a plate is cut nearly
perpendicular to the thickness axis, the cross-section is very irregular
in shape and contains pockets of untransformed austenite. Fig. 15 is
a higher magnification photomicrograph of a typical "plate". Note the
irregular lath-like interface which grows off at an angle of 25 to 300
from the midrib. Many times, this lath-like interface of the plates is
indistinguishable from the surrounding
lath-like morphology. The
twin-related loose-packets inFig. 14 probably do not emanate from a
midrib because the former usually occur with a much broader range of
angles relative to the twin plane, and because inspection of Fig. 15
reveals that the "laths" are parallel to each other on opposite sides
of the midrib and are not mirror-imaged as in Fig. 14. The linear
II-------
-
'
`
50
FIG. 14
AN EXAMPLE OF A TWIN-RELATED
IN Fe29.6Ni
TRANSFORMED WITH 60 kOe AT -2 0 C.
LOOSE -PACKET
91
FIG. 15 AN EXAMPLE OF THE TRANSITION PLATE
TRANSFORMED
MORPHOLOGY IN Fe29.6Ni
WITH A 9OkOe FIELD AT -2 0 C.
nature of the pockets of retained austenite and the straight a'- y
interfaces on some of the plates indicate that the slower growing outer
regions of the plates grow unidirectionally and in antiparallel directions
on either side of the midrib.
These observations suggest that plate-
thickening and lath-like martensite are both slow-growing slip martensites.
In light of the jagged nature of some plate interfaces and of lathlike martensites, we must be careful to distinguish between the jagged
continued growth of existing martensite and the "true" autocatalytic
nucleation of a martensite unit having a different growth direction than
its precursor.
According to the discussions in Section V-4.2, it is
highly improbable that the autocatalytic nuclei preexist as distinct
entities in the parent austenite.
Thus, there is a fine line between
continued growth of a given martensitic unit and autocatalytic nucleation.
Most of the martensite we see in this Fe Ni alloy is best thought
of as martensite cluster-spreading rather than the individual nucleation
events of separate embryos.
Davies and Magee (42 ) mapped out the observed structure of a series
of Fe Ni and Fe Ni Co alloys on a plot of ec
- Ms versus Ms . They
found that, for these alloys, austenite ferromagnetism above Ms is a
necessary, but not sufficient, condition for the formation of lenticular
rather than packet martensite.
The structures of the present zero-
field transformations in all three binary Fe Ni alloys fit into their
classification scheme.
However, as the magnetic field is applied, Ms in-
creases while ec increases much less; consequently ec - Ms decreases with
increasing field. According to the correlation plot, this should shift
the structures to more lath-like martensite when, in fact, the observed
structures are more plate-like in nature. Davies and Magee interpreted
their results in terms of austenite strength, i.e. above some critical
austenite flow stress, lenticular martensite would require less energy
to form than packet martensite because the latter includes slip in austenite
The inconsistencies of our field-
as an intrinsic part of its formation.
induced structures with their plot would have to be interpreted as a
decrease in austenite strength with increasing fields. There is some
evidence for this as will be discussed in Section V-3.4; however, we
will show in the next section that the lath-to-plate transition is best
correlated with other kinetic parameters.
V-3.3 Activation-Energy Correlations
Eq. 26 was used to calculate the activation energy, AWa, from T
0.3'
the time to transform to 0.3% martensite, Vi the average initial plate
volume, ni , the concentration of initial embryos, and v, the attempt
frequency. (This is only an approximate calculation, as will be discussed
in a later section).
The value of r0.
3
adopted for these calculations
was the median of several runs under identical testing conditions.
was measured using the procedure outlined in Section IV-6.3.
Vi
Because
of the lath-like nature of this morphology, the measurements were extremely difficult and time-consuming.
In addition, the thin-disk or
oblate-spheroid approximations does not work well for lath-like structures.
For these reasons, the value of Vi ( = 7 x 10- 9 cm3 ) is only an order of
magnitude value. Any inaccuracies in •i1 will just change all the AWA
values by a constant factor. Magee's experiments (22) suggest that ni
should be ~105 embryos/cc for small particles; however we felt a value
of 107 was more appropriate since we are testing the bulk kinetics of
a material that has been subjected to a complex thermal and mechanical
history. The attempt frequency was taken as 1014 sec - 1 to be consistent
with the Raghavan-Cohen growth path paper(50); however, using a value of
1011 sec - 1 as suggested by Magee(28)does not change the overall shape
of the curves.
The choice of these numbers is not very critical since
v.
AWa (from Eq. 26) is not a sensitive function of the product n.
1
The results of these calculations and the values of the chemical
and magnetic driving forces calculated from Eqs. 11 and 27 are listed in
Table 4 and plotted in Fig. 16.
The use of magnetic fields has allowed
us to extend the range of AWa values over that which was possible by
varying temperature alone. Note that the AWa values decrease somewhat
with increasing driving force or decreasing incubation time. This would
be due to a factor of 2 increase in initial plate volume with higher
driving forces, but it is more likely caused by autocatalytic effects
which are not taken into account in Eq. 26.
Since we have two independent variables - temperature and magnetic
field - which can change the driving force, we are in a position to
evaluate the relative effects of driving force and temperature.
When
the morphology and kinetics of two specimens with nearly the same
AWa or AgT are compared, the field-induced specimen is found to have
greatly accelerated kinetics compared to the lower-temperature specimens.
-
__
TABLE 4
Incubation Times, Activation Energies, and Driving Forces for Fe 29.6 Ni
Magnetic
Field
Chemical AG
Magnetic AGm
Total AgT
T0.3%
(k0e)
(cal/mole)
(cal/mole)
(10 9 erg/cm 3 )
(min)
AW
a
(10-12 ergs/event oK)
9
257.2
0
- 1.56
23.5
1.55
- 11
259.4
0
- 1.57
12.1
1.54
- 13
261.5
0
- 1.58
4.0
1.46
- 15
263.7
0
- 1.60
1.0
1.39
-
2
249.5
4.0
- 1.53
40.1
1.62
-
2
249.5
6.0
- 1.54
17.0
1.58
-
2
249.5
8.0
- 1.56
3.1
1.52
-
2
249.5
10.0
- 1.57
0.3
1.44
+
9
237.5
12.2
- 1.51
26.8
1.67
+
9
237.5
14.3
- 1.52
19.2
1.65
+
9
237.5
16.3
- 1.53
Temperature
(oC)
-
1.2
1.54
1.75
o ZERO-FIELD
t
03
1.65
-
L
CrJ
4-
U,
co
N
!m
0
o
<
1.55
1.45
D
.L
1.35
- 1.
I
I
62
___
I
-1.58
Ag T
FIG. 16
I
1
I
-1.54
( 109 ergs/cm 3 )
!
-1.50
ACTIVATION ENIERGY VERSUS TOTAL
GIBBS FREE -EN ERGY CHANGE FOR
Fe 29. 6 Ni (THE DASHED LINE CONNECTS
THE INITIAL LOWER DRIVING- FORCE
POINTS OF EACH SERIES.)
One might not have expected that two identical
specimens transforming
with the same activation energy or driving force would have different
kinetics and morphology.
The reason behind this behavior is that the
field not only accelerates the reaction by increasing the driving force,
but it also enables it to occur at higher temperatures where the reaction
reaps the additional benefit of higher thermal energy.
kinetic parameter is AWa/T which is plotted in Fig. 17.
The appropriate
When comparisons
are made between specimens with nearly the same AWa/T (and different
AgT), they have similar kinetics and morphology! The transition from
lath-like to plate-like occurs at roughly AWa/T = 5.4 x 10- 1 5 ergs/event
oK
in Fe 29.6 Ni.
This is at a region that is barely measurable because
of the small incubation times.
The preceding discussions are summarized graphically in Fig. 18;
higher rates of transformation yield plate-like martensite, whereas
slower rates yield a more lath-like structure. When this criterion is
applied to the Fe 28.7 Ni alloy, the measured transformation curves have
incubation times (or maximum transformation rates) that would fall in
the lath-like region in Fig. 18 in agreement with the observed morphology
illustrated in Fig. 8. The Fe 30.8 Ni alloy, to be discussed in the
next section can show a large incubation time followed by a very rapid
bursting region to plate martensite.
Thus, the rate of transformation
appears to be a better correlation parameter than the incubation time.
The calculation of AWa/T from Eq. 26 using the measured incubation time
is only a convenient approximation to the overall rate of transformation.
6.0
-
0
A-
o i
C
a)
U,
5.8
5.6
0
5.4
5.2
-1.62
-1.58
-1.50
= 1.54
Ag T( 109 erg s /cm
3
)
FIG. 17 ACTIVATION ENERGY DIVIDED BY THE
TEMPERATURE VERSUS TOTAL GIBBS
FREE -ENERGY CHANGE FOR Fe29.6 Ni
(THE SOLID LINES ILLUSTRATE THE FIT
WITH THE KCR MODEL.)
2
Id
-I
-2
-3
-4
5.3
5.1
AW
FIG. 18
5.5
/T , AU */T
LOGARITHM
5.7
6.1
5.9
(x 1O-15 ergs /event OK)
OF THE RATE OF
TRANSFORMATION VERSUS EXPERIMENTAL
AWo/T (CLOSED POINTS) AND CALCULATED
AU*/T (OPEN
POINTS), ILLUSTRATING THE
MORPHOLOGICAL TRANSITIONS IN Fe 29.6 Ni
(THE SOLID LINE IS FROM EQ.26)
100
To hope to find a simple relationship between driving-force and
temperature which will completely describe all the kinetic and morphological aspects of the martensitic transformation is a bit optimistic, to
say the least. The KCR model employs these parameters and others, and
still it requires one or two floating parameters.
Itis still true,
however, that reactions with lower activation energy-to-temperature
ratios tend to generate more plate-like products.
This conclusion
says that the commonly observed trend toward more plate-like structures
at lower temperatures (e.g. Fig. 12) is actually due to the fact that the
activation energy isvery much lower at the lower temperatures because
of the higher driving forces there. Therefore, itnow appears that low
temperature, at a given driving force, favors lath martensite and high
temperature favors plate martensite. The magnetic fields have enabled
us, for the first time, to separate the effects of driving force and
temperature.
The quantitative kinetic comparisons have been limited to the
incubation times. The overall kinetics of the transformation curves
have not been analyzed using Eqs. 12 and 13 because itwas felt that
these equations do not give an adequate description of the transformation
mode; therefore, the calculated parameter (AWa and p) from curvefitting would not have any real significance. The troublesome issues
are (i)the use of a single activation energy to describe both initial
and autocatalytic nucleation, (ii)the consideration of the autocatalytic embryo as a separate entity from the parent phase, and
(iii) the assumption of an autocatalytic factor, p, per unit volume of
transformed phase rather than per unit area of moving interface as the
101
discussions of Section V-4.2 imply. The development of the correct
equations to describe these kinetics was outside the objectives of this
thesis, but will be considered in the continuing program at MIT.
Fit with the KCR Model
V-3.4
The solid lines drawn in Figs. 16 and 17 illustrate the best
This fit looks very good in light of
possible fit with the KCR model.
the assumptions made in calculating AWa; however, the following section
describing the procedures and assumptions necessary to arrive at these
curves indicates that the model does not work as well as one might
suppose from Figs. 16 and 17.
First, Eqs. 15 to 17 are
The fitting procedure is as follows:
used to fit a point ( -90C, H = 0) on Fig. 16.
This is done by ad-
justing the radius and semithickness of the most potent embryo. We
assumed that the preexisting embryo is somewhere on the minimum energy
fall-line off the saddle point along the free-energy surface, given by
c /r = a/A. It was not necessary to assume this relation to satisfy
any conditions of the model; however, it provided a basis for the
choices
of re and ce. The re and ce values for an exact fit at the
O
O
chosen point were found to be 761A and 27A respectively. In addition to
substituting the density and shear-modulus values of Table 3 into Eqs.
15 to 17, we also assumed that A and a are proportional to the modulus,
as described previously. Using these procedures, the KCR model fell
short of fitting the -20C and +90C field-induced points.
Since the KCR model is very sensitive to the shear modulus,
i,
102
it was necessary to examine the effect of magnetic fields on the modulus.
Most reported modulus values increase with the application of a field
on the order of 2 kOe. (An increase of p with H would worsen the fit)
This, however, is not indicative of an intrinsic modulus change, but
rather is a special case of the well-known modulus defect phenomenon.
In short, when we measure the modulus of a demagnetized specimen, we
are also including a magnetoelastic strain due to domain reorientation.
When a field is applied, the domains become oriented with the field
and do not move appreciably with the application of a stress.
Thus,
modulus measurements under applied magnetic fields give values of the
moduli that are more characteristic of the elastic behavior of the
lattice.
We are interested in the variation of this intrinsic modulus
with high fields. Alers, et al.(l08)reported that 1
-3 x 10
-4
-4
C44
dc4 4
=
dH
at 00 C for an Fe 31.0 Ni alloy under fields in the range of
4 to 10 kOe.
This value shifted the KCR curves, matched to the -2°C
and +9°C data, in the correct direction, but it overshot the points.
The fit shown in Figs. 16 and 17 was obtained by assuming
I dH
-2 x 10-4 . This is a reasonable number, but it still must be regarded
as a floating parameter in the absence of a measured value.
modulus were
If the
not a function of temperature and field, then the three
lines of Fig. 16 would superimpose since AWa (i.e. AU*) is primarily
only a function of p and AgT in the KCR model.
lines of Fig. 17 would not superimpose
On the other hand, the
under these conditions; conse-
quently, this is a more revealing comparison with any model.
103
The values of the KCR critical loop radius, p*, in the region of the
isothermal transformations represented in Figs. 16 and 17 are near
5A.
Since this is less than twice the Burgers vector, core-overlap effects
which are not taken into account in the KCR model could be significant.
Fitting Fig. 17 is a severe test for any model.
The fit with the
KCR model is fair in light of the discussions of the last paragraphs;
however, it seems to fit best on the tailing-off points at higher driving forces. As pointed out earlier, these AWa values were probably
influenced by autocatalysis which is not accounted for in Eq. 26.
The
initial, lower driving-force points of each series are the most
representative of initial martensite nucleation. These are connected
by a dashed line in Fig. 16.
this line.
The KCR model does not match the slope of
Assuming that the size of the most potent embryo is a function
of the magnetic field does not improve the situation.
104
V-4
V-4.1
Fe 30.8 Ni -- (Bursting)
Kinetics
Fig. 19 illustrates the zero-field behavior of the Fe 30.8 Ni alloy.
The Ms temperature of - 45.2 0C was an average of 8 specimens which showed
a + 80C scatter.
Within this scatter band, the final percent martensite
was a linear function of Ms . The transformation curves are shown for
three temperatures; however measurements were also made at 40 K in order
to test the hypothesis
(16 )
that the bursting alloys are isothermal and
show a C-curve behavior (versus the KCR model prediction that a pure
athermal behavior exists at temperatures below which AU* = 0).
The 40 K
curve (not shown in Fig. 19) fell 0.5% below that at - 196 0 C. This may
indicate that the overall transformation behavior of the alloy exhibits
a C-curve behavior; but, 0.5% is a small difference and other complications could cloud the issue.
In addition, the driving force versus tem-
perature curves (shown in Fig. 5) demonstrate that, below about 100 0K,
there is little change of AGY ' with temperature; therefore, even an
athermal reaction would not benefit by a decrease in temperature.
Even
at 40K, however, audible clicks could be heard for about 12- minutes.
Again, this implies that the reaction is isothermal even at these very
low temperatures; unfortunately, the enthalpy change heating effects could
cause significant temperature fluctuations after the initial burst.
It
appears that this issue cannot be clearly resolved with bulk specimens,
but requires small-particle expreiments with bursting alloys.
At higher temperatures, we found positive proof that the reaction
105
Fe 30.8 Ni
NO APPLIED FIELD
100
mL
zLU1
80
- -196 0 C
140 0 C
(I)
H
z
LU
LU
0_
60
r-
40
E-
-4 5.2 0 C (=Ms)
20
0
FIG. 19
I·
I
I
·
20 40 60 80 100 120 140 160 180
TIME (MINUTES) --TRANSFORMATION KINETICS OF
Fe30.8Ni UNDER ZERO-FIELD
CONDITIONS.
10
proceeds isothermally even though this is a bursting alloy. The specimens were doped to prevent surface nucleation, then slowly cooled and held
for about 5 to 10 minutes at a series of decreasing temperatures just
above Ms . Incubation times as long as 1.5 minutes were observed followed
by a burst.
This is definite proof that the reaction is isothermal near
the Ms temperature.
Two specimens were impregnated with carbon by annealing 1 hour at
800 0 C in a vycor tube filled with CO.
The Ms temperature was suppressed
by 9°C and one specimen had a 6.8 minute incubation time prior to bursting.
This illustrates that interstitial pinning has a strong influence
on the initial nucleation event but not on the subsequent autocatalytic
bursting mode.
When a magnetic field was applied at - 200 C, which is
Ms , the curves of Fig. 20 were obtained.
',
25% above
The most obvious feature of thes
runs is that at 90 kOe a large (".30 minute) incubation time was observed.
0
The application of the KCR model to this alloy (assuming that re= 761 A as
in the Fe 29.6 Ni alloy and that the shear modulus values of Table 3 can
be extrapolated to lower temperatures) does predict that the - 200C/90 kOe
specimen should have a longer incubation time than the - 450C specimen
at zero field, but the relative magnitudes are a factor of 10 off. This
is not disturbing, however, in light of the approximations made.
This
exercise also predicts that the true zero-field Ms according to the KCR
definition (i.e. AWa = 0) is at
"
- 140 0C.
107
Fe 30.8 Ni
MAGNETIC FIELD APPLIED AT -20 0 C
100
i-- 80
m
140 kOe
,
H
90 kOe
,
60
I
z
I
I
I
I
I
40
LU
H
H
UJ
c
O
LU 20
I--
I
I
I
O0
00
v
FIG. 20
Ii
i
20
i
I
I
i
40 60 80 100
TIME (MINUTES
i
i
20 140 160 180
--
FIELD-INDUCED TRANSFORMATION
KINETICS OF Fe 30.8 Ni AT- 20 0 C.
108
V-4.2 Morphology and Autocatalytic Nucleation
The zero-field structure of this alloy is shown in the micrographs
of Fig. 21.
This is definitely a typical {259} bursting alloy.
Y
The zig-
zag cooperative growth of these martensitic plates is very much in evidence, especially when quenched to liquid-nitrogen temperatures.
When two
plates impinge, they do not just meet at the point of impingement but
rather they always meet at a common interface which can extend virtually
the whole length of the plate.
This illustrates the slow-growth stage of
the plate-thickening step.
These Fe 30.8 Ni martensitic plates have a much smoother a' - y inter-
face than those which were observed in the Fe 29.6 Ni alloys.
From the
discussion in Section V-3.3, it appears that the overriding factor governing the "degree-of-jaggedness" of the martensite-plate interface is its
ability to overcome barriers.
Under conditions of low AWa/T, the marten-
site interface breezes through obstacles very easily and grows at a fast
rate; the atoms probably move in a long-range cooperative military fashion
and are strongly influenced by such factors as mechanical constraints over
long distances. Under conditions of high AWa/T, the interface can be more
readily pinned at obstacles, thereby causing the lath-like effect at the
advancing plate interface.
The field-induced structure at - 200 C shown in Fig. 22 is, once
again, very similar to the zero-field structures at Ms . So far we have
shown that the magnetic field can influence the morphology by changing
the driving force or the temperature of transformation; however, there
has not been any apparent intrinsic effect of the field itself. As
109
(a)
5
O/A.----
(b)
FIG. 21
ZERO - FIEL D BURSTING
IN Fe 30.8 Ni i
(a) - 46 0 C
(b) - 1960 C
MORPHOLOGy
111
indicated in Section 1-2, we might expect some orientation dependence of
the plates with the field; however, it is not likely to find this effect
in an alloy, such as this, in which the cooperative growth of neighboring
plates is the overriding factor. This potential orientation effect will
be examined on a more quantitative basis in connection with the next alloy.
The Fe 30.8 Ni structure illustrates three of the important autocatalytic nucleation sites for bursting alloys.
The most obvious sites
are at the impingement of midribs on growing plates. This is particularly
evident at liquid-nitrogen temperatures, as shown in Fig. 21b.
These
autocatalytic sites must obviously be freshly created because it is highly
unlikely that pre-existing embryos exist that are ready to fire off in the
precise zig-zag array.
The second type of site, which must at least initially precede the
midrib-impingement type, occurs when the lattice-invariant deformation
mode during plate thickening changes from twinning to slip.
Operation of
this type of site probably generated the three plates shown in Fig. 23
which extend from the midrib.
Both of the previously mentioned sites are
always generated at an advancing martensite interface and are always associated with a prior midrib.
Christian(46) suggests that the interface determines the morphology,
whereas Owen, et al.
( 45 )
say that the growth rate is the important factor.
Both views are self-consistent with the AWa/T correlation, if the growth
rate determines the interface (or vice versa); this, in turn, could determine the morphology and the nature of the autocatalytic nucleation sites.
One very important autocatalytic site has been ignored up to now.
~rrr~:"~nvua*;_sr~pp-rr?-p~-_ --~-U~b~L~
112
20z
FIG. 23 AN EXAMPLE OF AUTOCATALYTIC PLATES
EMITTING FROM A MIDRIB IN Fe 30.8 Ni
TRANSFORMED UNDER ZERO-FIELD
CONDITIONS AT -46 0 C
113
The two sites just mentioned presumably operate mainly within a grain.
Without a mechanism for spreading into other grains, we would depend on
the initial nucleation sites to start up the transformation in the many
grains.
Estimates of 107 nuclei/cm 3 mean there is only about 1 nucleus for
every 2 grains, with a grain size of 0.04 mm; but it is apparent that almost all the grains contain martensite.
In addition, when an alloy such
as Fe 30.8 Ni transforms, every grain fills up in a matter of seconds.
The probability of every initial nuclei in every grain transforming at
the same time is miniscule.
Examination of specimens etched to reveal
the austenitic grain boundaries show that many of the plates and even more
of the midribs which hit a grain boundary are also touching a plate in the
next grain.
In addition, it was not uncommon to see midribs extending
through a grain boundary and bent by a small angle. The effectiveness of
the autocatalytic transfer from grain-to-grain is most likely a function
of the relative orientations of adjacent grains.
These observations in-
dicate that plate growth or autocatalytic mechanisms can and do operate
across incoherent grain boundaries.
It is conceivable that a midrib
moving at one-third the speed of sound in metal could shock-induce a new
plate at the point of contact; however, since we do not know the exact
nature of the incoherent grain boundary, we must speculate somewhat about
the other possible mechanisms involved.
Since slower growing lath-like martensite can also spread to other
grains, the midrib shock mode cannot be the only mechanism operating.
Careful re-examination of the Fe 29.6 Ni specimens reveals that the lathlike martensite often grows right along one side of a grain boundary.
114
In many of these cases, martensite also grows on the other side of the
grain boundary adjacent to the first martensite packet. This is most
apparent in specimens which have just begun to transform and are not complicated by a large quantity of martensite. One possibility is that the
grain boundary may have small regions coherent enough to let the slowgrowing martensite pass through to the next grain.
It was more common,
however, to see the spreading through grain boundaries in the cases where
the loose-packets hit a grain boundary head-on rather than in the packetthickening direction.
Thus, if the impinging shear is normal to the grain
boundary, the spreading occurs more effectively.
In any event, both of
these lath-like spreading modes were less evident than the impingingmidrib mode, i.e. there were more examples in which grain boundaries confined the lath-like reaction.
It follows that a bursting alloy has faster
kinetics both because of the cooperative nature of its growth and because
it has more efficient autocatalytic nucleation mechanisms.
115
V-5
V-5.1
Fe 22.5 Ni 4.0 Mn -- (Isothermal C-Curve)
Kinetics
The FeNiMn alloy studied here is at an exciting composition.
It will
not detectably transform by isothermal holding at any temperature.
How-
ever, by applying magnetic fields in the range of 60 to 140 kOe, we were
able to transform this alloy to up to 50% martensite. This illustrates
quite clearly that applying a field is very different from lowering the
temperature.
Fig. 24 demonstrates the effect of applying higher and high-
er fields at constant temperature.
Once again, as in the binary FeNi
alloys, the field accelerates the reaction; however, with the Mn alloy
there is no final plateau stage.
This is reminiscent of their differences
in the magnetization curves of Fig. 4.
It is obvious that there is an
intrinsic difference in the growth or autocatalytic spreading of these
two types of alloys.
Fig. 25 illustrates the effect of varying the temperature at constant field. The transformation is definitely isothermal over the whole
temperature range and exhibits a C-curve behavior, which is typical of
FeNiMn alloys.
The higher Mn content of the present alloy shifts the
C-curves out of the experimentally detectable range, until they are
"pulled back" by high magnetic fields.
The variation of the average time
for 0.3% martensite as a function of temperature and field is shown in
Fig. 26.
The mean values plotted in this figure indicate that the reac-
tion is definitely C-curve in nature with a nose temperature near - 120 0 C.
Attempts to further pinpoint the nose temperature, using interrupted
116
Fe 22.5 Ni 4.0 Mn
-81 oC
60
50
cl
z
40
30
zL
20
I0
n
0
FIG. 24
40 160 180
20 40 60 80 100 120 I1
TIME (MIN UTES)
FIELD-INDUCED TRANSFORMATION
KINETICS OF Fe 22.5 Ni 4.0 Mn
AT -81 OC.
117
Fe 22.5 Ni 4.0 Mn
140 kOe
60
50
tcn
H
z/)
z
40
30
LuJ
H
a: 20
u~J
C-)
a:
W
0_
10
0
O
FIG.25
20 40 60 80 100 120 140 160 180
TIME (MINUTES)
FIELD-INDUCED TRANSFORMATION
KINETICS OF Fe 22.5 Ni 4.0 Mn
AT 140 kOe AS FUNCTION OF
VARIOUS TESTING TEMPERATURES.
,-
* 140 kOe
118
A 120
10
00
90
80
-40
-80
I
U
H-
Ir
I.
•
A
-
.4-!
I
!
I 4-
I
-120
1p
-160
I
-200
.
0.01
0.I
1.0
i",,
10
I
L
I00
1000
( min.)
.
T0.3%
FIG.26
TIME TO TRANSFORM TO 0.3% MARTENSITE
AS A FUNCTION OF MAGNETIC FIELD
AND TEMPERATURE ( THE SOLID LINE
CONNECTS THE AVERAGE TIMES AT
EACH TEMPERATURE AT 140 kOe)
119
upquenching and downquenching experiments, will be discussed in Section
V-6.2.
The scatter involved in these experiments was met by repeating each
condition as many as 15 times.
Unfortunately, this used up so many speci-
mens and so much magnet time that a complete C-curve at lower fields was
not obtained.
The question of whether the C-curve shifts to shorter times
by moving horizontally along constant temperature lines or whether it also
shifts up or down in temperature is one of fundamental significance.
It
is difficult to draw any conclusions from the sparse low-field data of
Fig. 26.
However, Shih, Averbach, and Cohen(1 09 ) empirically correlated
the peak temperature versus % Mn.
They normalized the available data to
23% Ni by assuming 1% Mn is twice as effective as 1% Ni.
According to
their correlation, the nose temperature of our composition should be less
than - 140 0 C, approximately 200 C lower than the 140 kOe field-induced peak
temperature.
Thus, apparently the magnetic field shifted the C-curve to
shorter times and to higher temperatures.
This behavior will be compared
to that predicted by the KCR model in Section V-5.3.
V-5.2
Morphology and quantitative Metallography
The morphology of this alloy as a function of reaction temperature
at constant field and approximately the same martensite content is shown
in Figs. 27 a to d. This is similar to the structures observed by Pati
(110) in his Fe 24.2 Ni 3.0 Mn alloy. The structure goes through a
subtle
transition as the temperature is decreased, i.e. the martensitic units
become smoother and look more plate-like. The higher-temperature
120
(a)
-60 °C
~-------~
20CL
I1C~-~cr
(b) -81 °C
FIG. 27
MORPHOLOGY OF THE Fe22.5Ni 4.0 Mn
ALLOY TRANSFORMED WITH A 140 kOe
FIELD AT VARIOUS TEMPERATURE
121
(c)
-140
0C
2O0u.
(d) -1960 C
FIG. 27
(CONTINUED)
122
structures look almost lath-like and are reminiscent of the Fe 28.7 Ni and
Fe 29.6 Ni structures in Figs. 8 and 12.
Pati( I ll) examined the electron
microscopy of his alloy and found an internally twinned structure which
is not characteristic of lath martensite.
these are jagged plates.
It appears therefore, that
Comparisons of the structures obtained with
various fields shows that the field has no noticeable effect on the morphology over the range of 100 to 140 kOe (other than the fact that there
would not be any martensite at all without the field).
Pati (25 ) has investigated the effect of temperature and volume fraction martensite on V, the average plate volume in his Fe 24 Ni 3 Mn alloy.
He found that during the early stages of the transformation (i.e. < 1%
martensite), 9 was equal to 4 x 10-10 cm3 near the nose of his C-curve and
decreased to 1 x lO-10 cm3 at 600 C above or below the nose temperature.
At about 30% martensite, V of specimens reacted at all temperatures was
found to be equal to
"
1 x 10
-10 cm33.
Pati used linear extrapolations to
connect his data because of the absence of the critical measurements between 2 and 33% martensite.
Other workers (e .g.27) have reported that 9
remains constant over the range of 7 to 35% martensite. The latter results are consistent with our data, as illustrated in Fig. 28.
We ap-
proximated the morphology as oblate spheroids and measured their volume
using the procedures outlined in Section IV-6.3.
We see at once that 9
is initially a strong function of the martensite content, but after about
10% transformation it is approximately constant.
(The least-square fit
representing the initial behavior is 9 = 1.150 x 10- 7 - 9.420 x 10-9 x
%a'.)
123
12
I0
tO
co
0
I>
0
FIG. 28
I0
20
PERCENT
MARTENSITE
30
---
AVERAGE PLATE VOLUME FOR
Fe 22.5 Ni 4.0 Mn AS A FUNCTION
OF PERCENT MARTENSITE,
TEMPERATURE AND MAGNETICFIELD INTENSITY.
124
This behavior is consistent with the hypothesis that martensite can
apread by autocatalysis from clusters of transformed plates.
The initial
plates of each cluster is the largest, thereby giving a large initial V.
The rapid initial drop-off of V with percent martensite occurs as the
smaller austenite pockets around the initial plates begin to transform.
The relatively insensitive region after 10% martensite is apparently
caused by the spreading of the cluster into untransformed austenite grains.
It seems obvious that V must eventually decrease after the clusters spread
throughout the specimen, i.e. as the retained austenite pockets become
smaller and smaller.
If,however, the final transformation occurs by the
thickening of existing plates rather than by the nucleation of new units,
V may even increase during the later stages of the reaction.
Except for the fact that the V values at - 196 0 C are somewhat higher
than the others, there does not appear to be any strong dependences on
temperature or field.
This disagrees with Pati's data which show a high-
er initial plate volume near the peak temperature.
The reasons for this
difference and for the large differences in the absolute value of our
plate volumes compared to his are not known at this time.
The r/c ratios plotted in Fig. 29 are more scattered than the 9
values; however, they do show a decreasing trend with increasing percent
martensite. This could be due to plate thickening during the later stages
of the transformation.
In-situ transformation studies with the U. S.
Steel Company high-voltage electron microscope of Pati's alloy did reveal
evidence of plate thickening in discrete steps.
These steps are on too
fine a scale to be resolved with optical microscopy.
__
125
[T(C ), H (kOe)]
5
E-
[-196,120]
o [-196, 120]
S
-60o, 140]
.[-81,140]
o*[-40, 140]
81[-8-,120]
[-8 I,S140]
[-81, 100]
[ 0,[-81,140]
S[-196,140]
I0
PERCENT
FIG. 29
20
MARTENSITE
30
RADIUS-TO- SEMITHICKNESS RATIO FOR
Fe 22.5 Ni 4.0 Mn AS A FUNCTION OF
PERCENT MARTENSITE, TEMPERATUR E,
AND MAGNETIC FIELD INTENSITY.
126
Finally, we decided to test the hypothesis discussed in Section I-2
that magnetostatic-energy effects might cause some morphological dependence
on the field direction.
In particular, the plates might tend to line up
with the field and have a higher r/c ratio by being parallel to the field
direction than by being normal to it.
Two identical specimens were trans-
formed simultaneously at - 810 C in a 140 kOe field to about 13% martensite.
One was held parallel to the field direction and the other perpendicular
to it. The transformation curves of these specimens were almost identical,
indicating that polarization effects of the external field are negligible.
Visual examination of the structures of these two specimens (Fig. 30)
does not reveal any tendency for the martensitic plates to align with the
field.
In order to test for subtle variations which might not be detect-
able by ordinary visual examination, a polar grid was superimposed on two
2000X blow-ups of each specimen.
A line was drawn through each plate and
its angle with the specimen axis was put into one of 18 difference angular
sectors.
In addition, the r/c ratios of these plates were measured as a
function of this same angle. As Fig. 31 illustrates, there is no correlation with the field direction.
Thus, kinetic autocatalytic effects seem
to override any thermodynamically predicted orientation dependencies.
Fig. 31 does suggest a tendency for a greater number of plates (which
have higher r/c ratios) to lie 20 to 400 from the specimen axis independent
of the field direction.
This behavior could be due to quenching stresses
or it could be a consequence of the swaging operation used to produce the
final specimen size.
The specimens, however, were air cooled in a vycor
encapsulating tube which is not a severe quench, and so if this effect
127
(a)
SPECIMEN
AXIS
20.u
(b)
H
go
S.
FIG. 30
ORIENTATION DEPENDENCE OF THE
MORPHOLOGY OF Fe22.5Ni 4.OMn AS A
FUNCTION OF THE DIRECTION OF APPLIED
FIELD (TESTED AT -81 0 C , 140 kOe) .
(a) FIELD PERPENDICULAR TO
SPECIMEN AXIS.
(b) FIELD PARALLEL TO SPECIMEN AXIS.
DIRECTION
FIELD II TO AXIS
-- A-- FIELD 1_ TO AXIS
-
OF
12 8
SPECIMEN AXIS
12
L.
8
4
4
0
r/c RATIO
8
12
%
2
5
NUMBER OF PLATES
FIG. 31
DISTRIBUTION OF THE NUMBER OF PLATES
ALIGNED IN A SPECIFIC DIRECTION AND
THEIR RADIUS-TO- SEMI THICKNESS RATIOS
AS A FUNCTION OF THE DIRECTION OF THE
APPLIED MAGNETIC FIELD.
129
were due to some strain-induced artifact, it must have survived the recrystallization anneal.
It is conceivable that the number and perfection
of annealing twins (which have been suggested (28) as potent sites for
martensite nucleation) could vary depending upon the local swaging conditions.
Examination of our transformed specimens, however, did not reveal
any pronounced tendency to nucleate at annealing twins.
Still another
possibility is that the swaging operation causes a fiber texture which
could lead to some plate-orientation effects, since we have shown that
autocatalytic spreading through grain boundaries is a function of the
relative orientations of adjacent grains.
130
V-5.3
Fit with the KCR Model
Eq. 26 was used to calculate the activation energy, AW from T0.3%
a,
assuming Vi is constant and equal to 9.3 x 10
cm i.e. as extrapolated
from Fig. 28.
The predictions of the KCR model were compared to these data
using a procedure similar to that outlined in Section V-3.4.
(The best
0
fit was found for re = 534 A.)
The results of these calculations are pre-
sented in Table 5 and Fig. 32.
The AWa versus AgT plot of Fig. 32 is similar to that for Fe 29.6 Ni
in Fig. 16, except that the slope of the curve is smaller for the FeNiMn
alloy.
The match with the KCR model appears fairly good over the whole
range of driving forces; however, once again, the lower driving-force
points do not fit too well.
Fig. 33, i.e. AWa/T versus AgT , is a more crucial test of the data.
The KCR model does correctly predict a C-curve dependence, but the detailed fit with the data is not good.
- 138 0 C while that observed
is
The predicted nose temperature is
- 120 + 50 C. More seriously, the model
predicts that as H increases from 0 to 140 kOe, the nose temperature
should decrease from - 1270 to - 138 0 C; whereas as discussed in Section
V-5.1, the empirical correlations of Shih et al.(1
09)
indicate that it
increases n 200C.
The KCR model is still our only prototype and it does afford some
interesting predictions.
At very high fields, in the range of 300 to
400 kOe, the predicted nose temperature drops rapidly with increasing
field, as does AU*, the predicted activation energy for nucleation. At
fields near 485 kOe, the nose of the C-curve becomes athermal as defined
,
.-
L
TABLE 5
Incubation Times, Activation Energies, and Driving Forces for Fe 22.5 Ni 4.0 Mn
Temperatur e
(0C)
Magnetic
Field
(kOe)
Chemical Ag
Magnetic Agm
(cal/mole)
(cal/mole)
Total AgT
(10 9 ergs/cm3 )
Median T0.3%
AW
a
(min)
(10- 12 ergs/event 'K)
-
40
140
- 298
- 30.7
- 1.99
56.3
1.49
-
60
120
140
- 317
- 26.4
- 2.08
- 317
- 30.8
- 2.10
13.5
2.6
1.32
0.76
100
- 334
- 22.1
- 2.16
120
140
- 334
- 26.5
- 2.19
5.0
2.0
1.15
1.13
- 334
- 30.9
- 2.21
0.7
0.76
- 100
140
- 349
- 31.1
- 2.30
0.3
0.96
- 110
110
- 355
- 24.5
- 2.30
1.9
0.83
- 115
100
140
- 358
- 22.3
- 2.30
- 358
- 31.2
- 2.36
0.9
0.4
0.92
0.91
80
140
- 371
- 17.9
- 2.36
- 371
- 31.3
- 2.44
12.0
0.4
0.86
0.76
120
140
- 385
- 26.2
- 2.50
- 385
- 30.6
- 2.52
3.7
4.5
0.40
0.47
-
81
- 140
- 196
132
4 1.4
-
0
KCR MODEL
c
80 kOe
140 kOe
I
2
/4
C,
cvJ
r
0
0
*
D
0
-
•
0
0.8
0.6
L
40 kOe
0.4
-2.1
-2.3
As,- (x I0 9 ergs/cm 3 ) FIG. 32 COMPARISON OF THE RAG HAVAN -COHEN
MODEL TO THE ACTIVATION ENERGY
VERSUS DRIVING FORCE
133
6.8
QJI
I
I
A
6.4
0I
E-
0
KCR MODEL
G,)
6.0
--
0
5.6
0
< +-
kOe
<1
5.2
4.8
-2.6
kOe
E-
I
I
I
-2.4
- 2.2
Ag (x 10 9 ergs/cm3 )
-2.0
I
I
No
FIG. 33 COMPA RISON OF THE KCR MODEL TO
AWA/T
VERSUS DRIVING FORCE
134
by AU* = 0. With fields as high as 595 kOe, there is a predicted isothermal mode below the athermal nose of the C-curve.
At higher fields
still, the reaction is athermal to 0.20K. These predictions can explain
the results of Malinen, et al.(86) who applied a pulsed field of 360 to
400 kOe to Fe 23.6 Ni (3.3 and 3.6) Mn alloys which had zero-field isothermal nose temperatures of - 1000 and - 140%0 respectively.
They did
not find a field-induced C-curve but rather an athermal-like behavior,
i.e. the extent of transformation in both alloys rose monotonically with
decreasing temperature (as depicted in Fig. 2d in Section 11-2.1).
Thus,
although the detailed predictions of the model leave something to be desired, the overall predictions can explain many of the experimental observations.
135
V-6
V-6.1
Embryo Experiments
High-Temperature Magnetic Annealing
The classical high-pressure embryo experiments of Kaufman et al.(58)
lend credibility to the K-C version of the martensite embryo, i.e. a
mini-plate of martensite in the parent phase. Unfortunately, these experiments have never been repeated, while other high-pressure work ( 5 9 )
not specifically designed to examine the embryo effects, have been able
to explain the suppression of Ms while cooling under pressure by pure
thermodynamic arguments without any recourse to second-order effects from
changes in embryo potency.
A portion of our investigation involved experiments similar to those
of Kaufman et al. using a magnetic field instead of pressure to vary the
driving force during austenitizing. The simple reasoning behind these
experiments was that, if the embryos exist at high temperatures and if
they are mini-plates as envisioned by the KCR model, then they may be
ferromagnetic, and a magnetic field could increase their size just as the
high-pressure should decrease it.
The thermal-magnetic history of a typical experiment is illustrated
in Fig. 34.
The encapsulated specimens were heated in a magnetic field
to temperatures ranging from 400 to 7000 C, held for 15 minutes to 1 hour
in a high magnetic field, and then air-blast cooled to 100 to 200 0 C. They
were then removed from their encapsulating tubes and transformed under
zero-field conditions at subzero temperatures. The results of these runs
were compared with control runs made without any magnetic field during the
ar
a
600
I HOUR HOLD
•
500
7
7
#
•
////i///////t
/
I
/
f
##
o/
I
I
/
.7*
/
400
/
/
7,
7o,
i/
.7
I
/
300
T(
/
/
/
/
0 C)
/
/
.7
7I
200
/
/
100
100
kOe
FIELD OFF
FIELD ON
0
F
TRANSFORMATION
F
RUN
-100
TIME
FIG. 34 THERMAL -MAGNETIC HISTORY OF A TYPICAL
EMBRYO EXPERIMENT
__j
137
high-temperature cycle.
The temperature of the thermal-magnetic hold is critical.
Fig. 35.)
a + y
(Refer to
First of all, we would like to be above the equilibrium
+
y solvus line in order to avoid any complication from equilibrium
ferrite.
(Magee( 2 8) has shown that prior equilibrium ferrite suppresses
the subsequent martensitic transformation.)
In addition, we must be above
the "freeze-in temperature" which Kaufman and Cohen(19) envisioned to be
near 537 0C. E. Owen and Y. Liu(lll)
while studying FeNi equilibrium dia-
grams showed that at temperatures below 500 0C the bulk samples would not
come to equilibrium within feasible annealing times (e.g. 1 day to 500 0 C,
1 year at 3650C); of course, for an embryo we would not require equilibrium
of a bulk specimen, but rather of a very small volume aided by dislocation
motion.
Finally, other magnetic annealing experiments(e.g. 112) designed
to induce short-range alignment of Ni atoms with the field have shown a
freeze-in temperature of 4000 C.
We must also be at least 500C below the ferrite Curie temperature,
0ca-,
of the alloy under study in order to have a significant degree of
ferromagnetism in the embryo. Determination of the Curie temperature of
a non-equilibrium phase is a tricky problem. Bozorth (1) published a plot
of eca versus nickel content, but the data were taken at a time when martensitic phase transformations were not well understood; therefore, the
'
y reverse reaction casts a shadow of doubt on their results. Colling
(113), however, determined 0Oc for a series of FeNiCo alloys and
a linear
extrapolation of his data to 33% Ni falls right onto an extrapolation of
Bozorth'.s data.
The Weiss-Tauer (114 ) eca values (also shown in Fig. 35),
138
1200
1000
800
600
A nn
200
0
-200
0
20
60
40
wt % Ni
80
100
FIG. 35 EQUILIBR IUM DIAGRAAM ILLUSTRATING
THE CUR IE T EMPERATU RES AND
MARTE NSITIC-TRANSFORMATION
RANGE OF Fe Ni ALLOYS (FROM
REFER ENCES 113-116)
139
calculated from the average Bohr magneton number of the alloy by averaging
the nearest-neighbor interactions, are higher than the Bozorth data over
the whole alloy range.
It is apparent that there is only a small "window" in which a magnetic anneal could affect a pre-existing embryo.
The bulk of the magnetic
annealing experiments were done with the Fe 28.7 Ni alloy which has the
highest aca*
The annealing temperatures were varied from 400 to 7000 C in
order to be sure to hit the "window".
The results of these experiments can be summarized as negative.
If
anything, there was a 1 or 20 C depression of the reaction by the magnetic
austenitizing.
We also found that the subzero transformation rate was a
weak function of the severity of the air-blast cooling quench, i.e. slower
cooling in the range of n 1°C/sec depressed the transformation by about
20C. This effect will be discussed shortly.
To circumvent any complications of interstitial pinning of prior
embryos (which did not seem to interfere with the Kaufman pressure experiments even though they used higher C contents), we decided to recrystallize
the parent phase in a magnetic field to produce "virgin embryos".
To ac-
complish this, the Fe 29.6 Ni specimens were first quenched into liquid
nitrogen producing about 90% martensite. They were then heated to 7000 C
under a 100 kOe field, held for 15 minutes, and quenched to room temperature ( lo1C/sec).
The subsequent transformation did not show any enhance-
ment of the magnetically annealed specimens over those used for control
purposes.
Once again, however, the magnetically annealed specimens were
somewhat (1 to 20C) depressed.
This could be due to small differences in
140
in quenching rates as will be discussed shortly.
These experiments are not watertight, but they do not offer any support for the hypothesis that the martensitic embryo is ferromagnetic in
nature. Experiments are currently under way to test FeNiCo alloys which
have higher eca temperatures and to test bursting alloys and small particles; the latter should reflect the effects of initial embryos without
undue autocatalytic complications.
One interesting result of the recrystallization study was that these
specimens (with or without the magnetic field) were much more sensitive
to cooling rate (particularly in the temperature range below 3000 C) than
the specimens which were not freshly recrystallized. This suggests that
embryos are created during recrystallization or during the subsequent
cooling and can be stabilized by slow cooling or by annealing at room
temperature.
We also found that only a factor of 2 decrease in cooling
rate (% 1°C/sec) can depress the observable transformation from - 70C to
- 350 C! This depression could be due to interstitial pinning of virgin
embryos in the sensitive temperature range below 3000 C, or it is also
possible that embryos are actually created in the temperature range
below TO from existing heterogeneities or through quenching stresses.
Additional experiments are needed to sort out these issues.
141
V-6.2
Low-Temperature Magnetic Annealing and Cycling Experiments
Working on the presumption that the autocatalytic embryos exist as
separate entities from the martensitic plates (which we have shown to be
unlikely), we decided to apply the magnetic fields to influence these
embryos in a series of upquenching experiments.
Examination of Fig. 9
reveals that the zero-field transformation of Fe 29.6 Ni at - 110 C has a
fairly linear region from 40 to 100 minutes.
We up- and downquenched
from this curve to various temperatures and fields and then returned to
- 11"C, in order to study the effect on the subsequent transformation.
The resistivity change during a typical run is shown in Fig. 36.
In this
experiment, the transformation was first interrupted by an upquench to
20%C and a magnetic field of 80 kOe, during which the transformation progressed isothermally by an additional 1% martensite. The transformation
on the subsequent downquench to - ll1C exhibited a 15-minute incubation
time.
In order to eliminate interstitial pinning or other stabilization
effects as a viable explanation or this behavior, the specimen was upquenched once again for 15 minutes at + 200 C, but without an applied field.
Upon returning to - 110C, there was only a very small incubation time;
later experiments showed that this was probably due to a small amount of
transformation which occurs during the up- and downquenching operations.
This effect is not intrinsically related to the application of the magnetic field*, because we can also induce an incubation time by upquenching
The magnetic field simply provides a greater driving force for the transformation.
a
9
I0.0
9.8
E
z
cn
9.6
9.4
9.2
9.0
8.8
0
40
80
TIME (MIN.)
FIG. 36
160
120
N-
LOW TEMPERATURE TRANSFORMATION IN Fe 29.6 Ni
INTERRUPTED BY UPQUENCHING, WITH AND WITHOUT
A MAGNETIC FIELD
200
143
from - 11C to temperatures as low as + 20C without a magnetic field.
The subsequent - 110 C transformation is also delayed by the small amount
of additional transformation at the higher temperatures.
We have shown that this behavior is not due to any stabilization effect; however, it is consistent with our interpretation presented earlier
in Section V-4.2 that autocatalytic nucleation occurs directly from a
moving interface. The small amount of martensite that is produced at the
upquenching temperature appears to be due to the thickening of existing
martensitic units.
Since it is growing under conditions of higher AWa/T,
this could inhibit the degree of autocatalytic nucleation.
If we had
taken the more classical approach and hypothesized that the small transformation during the upquenching hold was due to the triggering off of the
most potent existing embryos, we would have to further hypothesize that
the new martensite did not produce any equally potent new embryos.
This
alternative does not seem to be in accord with the optical metallography
described in Sections V-3.2 and V-4.2.
Attempts were also made to pinpoint the nose of the C-curve in the
Fe 22.5 Ni 4.0 Mn alloy by altering the temperature of the transformation
during the course of the reaction.
These runs were not successful because
small amounts of martensite formed during the up- and downquenching and
inhibited the subsequent transformation, as in the Fe 29.6 Ni case.
The
correlations that could be made, however, indicated that the nose of the
C-curve determined in this manner (> - 110C) was higher than that indicated from the incubation times (- 120 + 50 C).
partitioned and stressed austenite
This illustrates that the
behaves differently from virgin
144
austenite.
Finally, we attempted to discover if the nose temperature of the
FeNiMn alloy changed with the field by increasing the field during the
course of the reaction from 110 to 140 kOe.
By repeating this type of
experiment at a series of temperatures, the ratios of the relative transformation rates before and after the field-increase should indicate
whether the C-curve is shifting with field.
These experiments were aban-
doned when it was noticed that the effect is not reversible, i.e. the
ratio of rates on decreasing the field is much greater than the inverse
of this ratio on increasing the field. This irreversibility casts some
uncertainty on the Peters, Bolton, and Miodownik(96) experiments which
were based on the interpretation of the increase of the transformation
rate on the application of a magnetic field.
145
VI.
CONCLUSIONS
1. Magnetic fields accelerate the martensitic transformation in FeNi
alloys both by increasing the Gibbs free-energy difference between
the product and parent phases and by allowing the reaction to occur
at higher temperatures.
2. The orientation of martensite plates and their radius-to-semithickness
ratios are not metallographically correlative with the direction of
the applied field.
3. Plate and lath martensites in FeNi alloys are two extremes of a continuous spectrum of morphologies.
The transition structures contain
a mixture of lath-like and plate-like martensite, each of which, in
themselves, are intermediate transition morphologies. The jagged,
lath-like interface of the transition plates is relatively slow-growing
slipped martensite which propagates in antiparallel directions 25 to
30" on either side of the midrib.
4. The initial nucleation of bursting alloys near their Ms temperatures
is thermally activated as evidenced by an observable nucleation time
which is increased by carbon additions.
5. The lath-to-plate morphological transition in FeNi alloys is not a
simple function of the temperature, Gibbs free-energy change, nickel
content, ecy - M
s , or the activation energy, AWa.
The best correla-
tion parameter is the rate of transformation which can be approximated
by the incubation time; the latter correlates with AWa/T.
As AWa/T
146
decreases, the morphology changes from lath to a lath-plate transition
and then to plate.
Lower temperature favors lath martensite which,
before the use of magnetic fields, has always been overshadowed by
the large increases in the chemical driving force with decreasing temperature.
An increase in driving force favors the plate morphology.
6. Autocatalytic nucleation sites in Fe-Ni alloys appear to be an integral part of the moving austenite-martensite interface. Three important sites are pointed out and discussed:
midrib impingement on
a growing martensitic plate, the transition region in which the
lattice-invariant mode changes from twinning to slip, and grainboundary impingement.
7. The Fe 22.5 Ni 4.0 Mn alloy does not transform detectably at any temperature until fields in the range of 60 to 140 kOe are applied.
The
transformation is then characterized by isothermal C-curve kinetics
which shifts to shorter times as the field is increased.
8. The effects of magnetic fields on the martensitic kinetic behavior
imposes a severe test on the current nucleation theories.
The Cohen-
Raghavan model can be fitted to much of the data; however, it does not
adequately explain the detailed kinetic behavior under high fields.
9. The average plate volume of the Fe 22.5 Ni 4.0 Mn alloy is insensitive
to temperature and magnetic field; however, it decreases rapidly with
the extent of transformation to 7% martensite and, thereafter, becomes
147
insensitive to the amount of martensite.
This behavior is correlated
with the autocatalytic spreading of clusters of martensitic plates.
10.
High-temperature magnetic-annealing experiments do not support the
hypothesis that the pre-existing martensitic embryos are ferromagnetic in nature.
11.
The Fullman quantitative metallography equations are extended to the
case of oblate spheroids.
148
VII
SUGGESTIONS FOR FUTURE WORK
1. Use electron microscopy to study the lath-to-plate transition on a
finer scale and to examine the details of the proposed autocatalytic
mechanisms.
2.
Use FeNiCo alloys, bursting alloys, and small particles to extend the
magnetic annealing experiments.
3.
Produce embryos by prior plastic deformation and use the magnetic
field to effect their potency.
4.
Devise magnetic mechanical-deformation experiments to show whether
this can be a viable commercial process for the fabrication of sheet
metal auto-body parts.
5.
Measure the incubation times of the Fe 30.8 Ni bursting alloy as a
function of temperature, magnetic field, and carbon content.
6. Use magnetic fields to induce martensite in Fe > 34 wt% Ni to test
the hypothesis that the transformation should be C-curve in nature.
7. Use single crystals for precision orientation effects of the magnetic
field.
8. Extend the quantitative metallography to determine the nucleation rate
versus percent martensite in magnetic fields.
9. Develop kinetic equations to fit the full transformation curves,
thereby determining AWa and the effect of the magnetic field on the
autocatalytic factor.
149
APPENDIX I
MAGNETIC CLAUSIUS--CLAPEYRON EQUATION
TO is the temperature at which the free energies of the austenite and
martensite are equal as expressed by
9, =gy
In order to maintain this equality for incremental temperature and magnetic-field changes (at constant pressure and composition), the following
relations must hold:
dg , = dg
'•g)-
dT +
3g
T xpdT + aH
dH =
p1
x,p
ag
aT
Y
x,
3gy•
pdT + DH
H x,p
dH
- S dT + I dH = - S dT + I dH
a
a
Y
dT_
I -I
dH
S -S
y-
Y
(A-l)
Y
I ASdT = f AIdH
(A-2)
If we assume that the heat-capacity difference at constant pressure is
not a function of temperature, (i.e. the slope of the free-energy curve
versus temperature is constant) and that the difference in magnetization
of the phases is not a function of the field, then Eq. A-2 can be
150
integrated:
T
H
dT = AI f
AS = I
TO
dH
0
(TO-T) = AT = HA-
(A-3)
in which TO is the equilibrium temperature corresponding to zero applied
field.
(This is the equation first derived by Meyer and Taglang(71) by
assuming that one can just equate the expressions for g , and gy and later
by Satyanarayan, Eliasz, and Miodownik
linear Ag versus T curve.)
(92 )
using similar triangles on a
For reversible equilibrium:
AS = -
TO
where L is the latent heat of the reaction or the enthalpy difference between the two phases.
Eq. A-3 can now be written as:
TOHAI
AT =
(A-4)
When applying this equation to the "athermal" martensitic transformation,
it is a useful approximation to set TO = MS (rather than the true TO in
which ga = gy) and to assume that surface and strain energies are integral
parts of the a' phase.
151
APPENDIX 2
QUANTITATIVE METALLOGRAPHY OF THIN OBLATE SPHEROIDS
Fullman(1
17)
has developed appropriate equations for the quantitative
metallography of multi-sized thin disks in a matrix.
In this appendix,
the analysis is extended to include thin oblate spheroids as a closer approximation to the martensite morphology than the previously-assumed thin
disks. Most of Fullman's thin-disk equations are actually better approximations for thin oblate spheroids because of the rounded nature of the
spheroidal rim.
Rather than duplicate the Fullman treatment, the reader is referred
to that paper(117) for all definitions and for the details of much of the
following development.
The equations for the thin oblate spheroid will
simply be listed unless they are significantly different from the Fullman
thin-disk analysis.
A.1
Particles of Uniform Size
Consider a sample containing particles of a phase a in the form of
thin oblate spheroids of radius r and semithickness c, where r >> c.
(c = t/2 in the Fullman notation.)
4
2
-
f = NvV = Nv( 4-r c) = Nss
=8s
=
x
rN s
rc
where s = average area of particles intersected by a random cross-section
152
2 4
4
f = NvV = NV-r c) = NLr = NV
2-
r k
S-c
3
where k = mean traverse length for plates of uniform size.
A.2
Particles of Nonuniform Size
When the particles under consideration are in the form of thin oblate
spheroids with a distribution of radius r and semithickness c, the distribution may be divided into a number of arbitrarily small ranges of particles of radius r i
,
and within each such range the plates may be further
subdivided into arbitrarily small ranges of semithickness cj.
Nsij -2 r.N
1 vij
sij
8
3y ri cj
Lij
2 r
ij
8
3 C
-
2V
Trr
4V
vij
153
The expression for E (average of the reciprocal lengths of particle
intersections with a random sectioning plane) is identical to the thindisk case:
4r
However, F for the oblate spheroids (average of the reciprocal thickness
of particle intersections with a random sectioning plane) is now a function of both 4 and X, as shown below:
F.
sin ý
=
2c.J1 - X2/r2
ij
r.
0 F P
rr/2
0
dXd
r.
1/2 sin b
f
f
0
0
2c.Jl - X2/r2
2
3cj
1 Nsi
ij
i ijNsi j
-i
2cj_
c
Nvijr
E21v
Ej
riij
= 2(r/c)
3r
Nvijri
4 sin
2
dedX
154
6 (average thickness-to-length ratios of the particle intersections with
a random sectioning plane) is derived as follows:
c.
G..
sin
r.
1
ij
71/2
G.i
=
f
c.
pdp
dý
0
r i sin
Tr/2
4 sin2
cj
-s
- sin d
ri sin 9
0
4c.
1
c
-N vij rir
ij2
4E=
7r
r
These equations may be combined to yield the following useful relations:
r4
4P
c
16
16
155
2-
2
8EN s
E
7r
2
NL
2EN S
N - f - 8fE - 8ENs
V
V
2
TS
r _ 373F
8i
(-)
=
2
156
BIBLIOGRAPHY
1.
Richard M. Bozorth, Ferromagnetism. New York: Van Nostrand
1951.
2.
Herbert B. Callen, Thermodynamics.
Sons, Inc., 1960.
3.
E. A. Guggenheim, Thermodynamics. New York:
Sons, Inc., 1967.
4.
E. A. Guggenheim, "On Magnetic and Electrostatic Energy", Proc.
Royal Soc. Series A 155 49(1936); "The Thermodynamics of
Magnetization." 70.
5.
S. Chikazumi, Physics of Magnetism.
Inc., 1964.
6.
D. J. Craik, Structure and Properties of Magnetic Materials.
London: J. W. Arrowsmith Limited, 1971.
7.
B. D. Cullity, Introduction to Magnetic Materials.
Addison-Wesley, 1972.
8.
C. Killel and J. K. Galt, "Ferromagnetic Domain Theory", Solid State
Physics 3 437 (1956).
9.
J. Goddard and J. G. Wright, "The Effect of Solution pH and
Applied Magnetic Field on the Electrodeposition of Thin
Single-Crystal Films of Cobalt", Brit. J. Appl. Phys. 15
807(1964).
New York: John Wiley and
N.Y.:
John Wiley and
John Wiley and Sons,
Reading, Mass.:
10. M. L. Bernshteyn, G. I. Gronik, and P. R. Dolzhanskiy, "Effect
of a Magnetic Field on the Phase Transformations in Nickel
Steels", Physics Metals Metallogr. 19 (6)77(1965).
11.
A. S. Yermolayev, A. Z. Men'shikov, and P. A. Malinen, "Development
of Uniaxial Magnetic Anisotropy on the Martensitic Transformation
in a Magnetic Field", Physics Metals Metallogr. 26 (1)73(1968).
12.
E. C. Stoner, "The Demagnetizing Factors for Ellipsoids", Phil.
Mag. 36 (7)803(1945).
13.
"Symposium on the Formation of Martensite in Iron Alloys", Met.
Trans. 2 2327-2462 (1971).
14.
Phase Transformations. Metals Park, Ohio: American Society for
Metals, pp. 1-180 (1970).
157
15.
V. Raghavan and A. R. Entwisle, "Isothermal Martensite Kinetics
in Iron Alloys", Iron and Steel Institute Special Report No.
93, p. 30 (1965)
16.
C. L. Magee, Transformation Kinetics, Microplasticity, and Aging
of Martensite, Ph.D. Thesis, Carnegie Institute of Technology,
(1966).
17.
V. Raghavan and M. Cohen, "Measurement and Interpretation of Isothermal-Martensitic Kinetics", Met. Trans. 2 2409 (1971).
18.
J. Philibert and C. Cussard, "Kinetics of the Martensite Transformation in a Hyper-Eutectoid Steel", J. Iron and Steel Inst.
180 34 (1955).
19.
M. Cohen, "Nucleation of Solid-State Transformations", Trans.
AIME 212 171 (1958).
20.
V. Raghavan and M. Cohen, "A Nucleation Model for Martensitic
Transformations in Iron-Base Alloys", Acta Met. 20 333 (1972).
21.
E. S. Machlin and M. Cohen, "Isothermal Mode of the Martensitic
Transformation", Trans. TMS-AIME 194 489 (1952).
22.
C. Magee, "The Kinetics of Martensite Formation in Small Particles",
Met. Trans. 2 2419 (1971).
23.
S. R. Pati and M. Cohen, "Nucleation of the Isothermal Martensitic
Transformation", Acta Met. 17 189 (1969).
24.
J. C. Fisher, J. H. Hollomon, and D. Turnbull, "Kinetics of the
Austenite-Martensite Transformation", Trans. AIME 185 691 (1949).
25.
S. R. Pati and M. Cohen, "Kinetics of Isothermal Martensitic Transformations in an Iron-Nickel-Manganese Alloy", Acta Met.
19 1327 (1971).
26.
V. Raghavan, "Formation Sequence of Plates in Isothermal Martensite
Transformation", Acta Met. 17 1299 (1969).
27.
D. G. McMurtrie and C. L. Magee, "The Average Volume of Martensite
Plates During Transformation", Met. Trans. 1 3185 (1970).
28. C. L. Magee, "The Nucleation of Martensite," Phase Transformations.
Metals Park, Ohio: American Society for Metals, p.115 (1970).
29.
J. F. Breedis and W. D. Robertson, "Martensitic Transformation and
Plastic Deformation in Iron Alloy Single Crystals", Acta Met.
11 547 (1963).
30.
J. F. Breedis, "Martensitic Transformations in Iron-Chromium-Nickel
Alloys", Trans. AIME 230 1583 (1964).
158
31.
G. R. Speich, "Tempering of Low-Carbon Martensite", Trans. AIME
245 2553 (1969).
32.
A. R. Marder and G. Krauss, "The Formation of Low-Carbon Martensite
in Fe-C Alloys", Trans. ASM 62 957 (1969).
33.
G. Krauss and A. R. Marder, "The Morphology of Martensite in Iron
Alloys", Met.Trans. 2 2343 (1971).
34.
T. Bell and R. G. Bryans, "The Effects of Prior Transformation and
Prestrain on the Habit Planes of Acicular Iron-Nickel Martensite",
J. Met. Sci. 5 135 (1971).
35.
J. S. Bowles and D. P. Dunne, "The Role of Plastic Accomodation
in the (225) Martensite Transformation", Acta Met.17 677 (1969).
36.
J. C. Bokros and E. R. Parker, "The Mechanism of the Martensite
Burst Transformation in Fe-Ni Single Crystals", Acta Met.
11 1291 (1963).
37.
G. Krauss and W. Pitsch, "The Fine Structure and Habit Planes of
Martensite in an Fe-33 Wt Pct Ni Single Crystal", Trans. AIME
233 919 (1965).
38.
R. L. Patterson and C. M. Wayman, "The Crystallography and Growth
of Partially-Twinned Martensite Plates in Fe-Ni Alloys," Acta
Met. 14 347 (1966).
39.
R. P. Reed, "The Plate-Like Martensite Transformation in Fe-Ni
Alloys", Acta Met. 15 1287 (1967).
40.
R. Huizing and J. A. Klostermann, "The Martensite Transformation in
Small (0.1-0.3mm) Iron-Nickel Single Crystals", Acta Met. 14
1693 (1966).
41.
W. S. Owen, E. A. Wilson, and T. Bell, High Strength Materials,
ed. V. F. Zackay, New York: John Wiley and Sons, p.167 (1965).
42. R. G. Davies and C. L. Magee, "Austenite Ferromagnetism and
Martensite Morphology", Met. Trans. 1 2927 (1970).
43.
H. Schumann, Arch Eisenhuettenw. 8 647 (1967).
44. G. Thomas "Electron Microscopy Investigations of Ferrous Martensites",
Met.Trans. 2 2373 (1971).
45.
W. S. Owen, F. J. Schoen, and G. R. Srinivasan, "The Growth of
a Plate of Martensite", Phase Transformations. Metals Park,
Ohio: American Society for Metals, p. 157 (1970).
159
46. J. W. Christian, "Military Transformations: An Introductory Survey",
Iron and Steel Institute Special Report No. 93, p.3 (1965).
47. T. Bell and W. S. Owen, "The Thermodynamics of the Martensite
Transformation in Iron-Carbon and Iron-Nitrogen", Trans, AIME
239 1940 (1967).
48. J. S. Pascover and S. V. Radcliffe, "Athermal Transformations
in the Iron-Chromium System", Trans. TMS-AIME 242 673 (1968).
49.
G. Olson, private communications.
50. V. Raghavan and M. Cohen, "Growth Path of a Martensite Particle",
Acta Met. 20 779 (1972).
51.
H. Knapp and U. Dehlinger, "Mechanism and Kinetics of the AusteniteMartensite Transformation", Acta Met.4 289 (1956).
52.
F. C. Frank, "Martensite", Acta Met.1 15 (1953).
53. J. W. Christian, "Martensitic Transformations: A Current Assessment",
Iron and Steel Institute Special Report No. 33, p. 129 (1969).
54.
J. W. Christian, private communications.
55. J. A. Venables, "The Martensite Transformation in Stainless Steel,"
Phil. Mag. 7 32 (1962).
56.
G. B. Olson and M. Cohen, "A Mechanism for the Strain-Induced
Nucleation of Martensitic Transformations", J. Less-Common
Metals 28 107 (1972).
57.
R. E. Cech and D. Turnbull, "Heterogeneous Nucleation of the
Martensite Transformation", Trans.TMS-AIME 206 124 (1956).
58.
L. Kaufman, A. Leyenaar, and J. S. Harvey, Progress in Very High
Pressure Research, Eds. F. P. Bundy, W. R. Hibbard, and H.
Strong, New York: John Wiley and Sons, Inc. p.90 (1960).
59.
S. V. Radcliffe and M. Schatz, "The Effect of High Pressure on the
Martensitic Reaction in Iron-Carbon Alloys", Acta Met. 10
201 (1962).
60.
M. Takahaski and T. Kono,
61.
C. D. Graham, Jr., J. Phys. Soc. Japan 16 1481 (1961); Suppl.
B-I 17 321 (1962).
62.
T. Sambongi and T. Mitui, J. Phys. Soc. Japan 16 1478 (1961);
18 1253 (1963).
J. Phys. Soc. Japan 15 936 (1960).
160
63.
R. Smoluchowski and R. W. Turner, J. Appl. Phys. 20 745 (1949);
Physics 16 397 (1950).
64.
B. Sawyer and R. Smoluchowski, J. Appl. Phys. 28 1069 (1957).
65. A. H. Geisler, J. P. Martin, E. Both, and J. H. Crede, Trans.
AIME 203 985 (1955).
66. V. S. Bhandary and B. D. Cullity, Trans. TMS-AIME 224 1194 (1962).
67.
0. L. Boothby, D. H. Wenny, and E. E. Thomas, J. Appl. Phys.
29 353 (1958).
68.
S. Chikazumi and C. D. Graham, Jr. "Directional Order", Magnetism
and Metallurgy. 2 Eds. A. E. Berkowitz and E. Kneller, N. Y.:
Academic Press, p.577 (1969).
69.
L. Neel, J. Pauleve, R. Pauthenet, J. Laughier, and D. Dautreppe,
"Magnetic Properties of an Iron-Nickel Single Crystal Ordered
by Neutron Bombardment," J. Appl. Phys. 35 873 (1964).
70.
C. Kittel, E. A. Nesbitt, and W. Shockley, "Theory of Magnetic
Properties and Nucleation in Alnico V", Phys. Rev. Series 2
77 839 (1950).
71.
A. P. J. Meyer and P. Taglang, "Influence of the Magnetic Field
on a Polymorphic Transformation of a Ferromagnetic Material",
J. Phys. Radium 14 82 (1953).
72.
H. Jahn, " The Influence of Heat Treatment in a Magnetic Field
on the Structure and Mechanical Properties of Steel", Stahl
u. Eisen 78 (3) 178 (1958).
73.
J. W. Cahn, "Magnetic Aging of Spinodal Alloys", J. Appl. Phys.
34 (12) 3581 (1963).
74.
E. G. Herbert, Iron and Steel Inst. 120
75.
V. D. Sadovskiy, N. M. Rudigin, L. V. Smirnov,G. M. Filonchik,
and I. G. Fakidov, "The Question of the Influence of Magnetic
Fields on Martensitic Transformation in Steel", Physics Metals
Metallogr. 12 (2) 131 (1961).
76.
E. A. Zavadskii and Fukidov, "Magnetization of the Intermetallic
Compound MnAu in Superstrong Pulsating Magnetic Fields ",
Physics Metalý Metallogr. 12 (6)47 (1961).
77.
Ye. A. Fokina and E. A. Zavadskiy, "Effect of a Magnetic Field on
the Martensitic Transformation in Steel", Physics Metals
Metallogr. 16 (2) 128 (1963).
239 (1929).
161
I
78.
M. A. Krivoglaz and V. D. Sadovskiy, "Effect of Strong Magnetic
Fields on Phase Transformationg',Physics Metals Metallogr.
18 (4)23 (1964).
79.
Ye. A. Fokina, L. V. Smirnov, and V. D. Sadovskiy, "Effect of
a Pulsating Magnetic Field on the Position of the Martensitic
Transformation Range in Steel", Physics Metals Metallogr. 19
(4)101 (1965).
80.
I. G. Fukidov, L. D. Voronchikhin, E. A. Zavadskiy, and A. M.
Burkhanov, "The Variation in the Magnetic Intensity of Austenitic
Steel Exposed to a Strong Pulsating Magnetic Field", Physics
Metals Metallogr. 19 (6)49 (1965).
81.
Ye. A. Fokina, L. V. Smirnov, and V. D. Sadovskiy, "Destabilization
of Austenite By a Strong Pulsating Magnetic Field", Physics
Metals Metallogr. 19 (5)73 (1965).
82.
E. I. Estrin, "Effect of a Magnetic Field on the Martensitic
Transformation", Physics Metals Metallogr. 19 (6)117 (1965).
83. Ye. A. Fokina, L. V. Smirnov, V. D. Sadovskiy, and A. F. Peikul,
"Effect of a Permanent Magnetic Field on the Martensitic Transformation in Steel", Physics Metals Metallogr. 19 (6)121 (1965).
84.
L. V. Voronshikhin and I. G. Fakidov, "Determining the Latent Heat
of the Martensitic Transformation Induced by a Magnetic Field
in Steel", Physics Metals Metallogr. 21 (3)119 (1966).
85.
P. A. Malinen and V. D. Sadovskiy, "Effect of a Magnetic Field on
the a + y Transformation in Iron-Nickel Alloys". Physics Metals
Metallogr. 21 (5)139 (1966).
86.
P. A. Malinen, V. D. Sadovskiy, and I. P. Sorokin, "Effect of Mag netic Field on the Martensite Transformation in Alloys of the 24
Ni 4Mn Type", Physics Metals Metallogr. 24 (2) 101 (1967).
87.
P. A. Malinen, V. D. Sadovskiy, L. V. Smirnov, and Ye. A. Fokina,
"Reasons for the Influence of a Pulsating Magnetic Field on
Martensitic Transformations in Steels and Alloys", Physics
Metals Metallogr. 23 (3)138 (1967).
88.
V. D. Sadovskiy, L. V. Smirnov, Ye. A. Fokina, P. A. Malinen, and
I. P. Sorokin, "Quenching Steel in a Magnetic Field", Physics
Metals Metallogr. 24 (5)144 (1967).
89.
G. I. Granik, M. L. Bernshteyn, and 0. D. Dolgunovskaya, "Influence
of Magnetic Field on Certain Phase Transformations in Nickel
Systems", Physics Metals Metallogr. 24 (6)151 (1967).
r
r
162
90.
L. D. Voronchikhin, L. N. Romashev, and I. G. Fakidov, "Influence
of the Initial State of Austenitic Steel on the Martensitic
Transformation in a Strong Magnetic Field", Physics Metals
Metallogr. 26 (5)183 (1968).
91.
L. N. Romashev, I. G. Fakodov, and L. D. Voronchikhin, "Influence
of a Strong Magnetic Field on the Martensitic Transformation
of an Iron-Nickel Alloy", Physics Metals Metallogr. 25 (6)
(1968).
92.
K. R. Satyanarayan, W. Eliasz, and A. P. Miodownik, "The Effect of
a Magnetic Field on the Martensite Transformation in Steels",
Acta Met. 16 877 (1968).
93.
Ye. A. Fokina, L. V. Smirnov, and V. D. Sadovskiy, "Effect of
Magnetic Field on the Position of the Martensitic Transformation Point in Carbon Steels", Physics Metals Metallogr.
27 (4) 190 (1969).
94. P. A. Malinen and V. D. Sadovskiy, "Effect of Magnetic Field on
the E c Transformation in Fe Mn Alloys", Physics Metals
Metallogr. 28 (6)54 (1969).
95.
K. R. Satyanarayan and A. P. Miodownik, "The Effect of Magnetic
Fields on Transformations in Steels", Institute of Metals,
Monograph and Report Series No. 33, 162 (1969).
96.
C. T. Peters, P. Bolton, and A. P. Miodownik, "The Effect of
Magnetic Fields on Isothermal Martensitic Transformations",
Acta Met. 20 881 (1972).
97.
F. D. Miroshnichenko, A. I. Stolyarenko, V. G. Kanibolotsky, V.
L. Snezhnoy, and A. M. Brynza, "The Effect of Pulsed Magnetic
Field on Phase Transformations in Steel khl5N5D2T at 400650 0C", Met. Sci. Heat Treat. 13 (13) 326 (1971).
98.
M. I. Prudnikov and A. P. Feygin, Kuznets.
5 6 (1964).
no-shtamp. proizv
99. Ye. A. Popov, Yu. A. Bochvar, S. M. Polyak, A. S. Stolbunov, and
D. V. Raykh, Kuznechno-shtampov. proizv. 5 1 (1966).
100. A. P. J. Meyer and P. Taglang, "Influence of the Magnetic Field
on a Polymorphic Transformation of a Ferromagnetic Material",
J. Phys. Radium 14 82 (1953).
101. B. S. Lement, B. L. Averbach and M. Cohen, "Microstructural Changes
on Tempering Iron-Carbon Alloys", Trans. ASM 46 851 (1954).
163
102. C. H. Shih, B. L. Averbach and M. Cohen, "Some Characteristics
of the Isothermal Martensitic Transformationr', Trans. AIME
203 183 (1955).
103. J. W. Christian, "Accommodation Strains in Martensite Formation
and the Use of a Dilatalion Parameter", Acta Met., 6 377 (1958).
104. A. J. Goldman and W. D. Robertson, "Elastic Properties of Austenite
and Martensite in Iron-Nickel Alloys", Acta Met. 12 1265 (1964).
105.
106.
107.
108.
109.
L. H. Bennett and L. J. Swartzendruber, "On the Interpretation of
Mossbauer Effect Spectra as Related to the Constitution of
Cu-Ni-Fe Alloys", Acta Met. 18 485 (1970).
A. P. Miodownik, "The Effect of Two Gamma States on the Magnetic
Properties of Fcc Iron-Cobalt Alloys", Scripta Met. 3 931
(1969).
L. Kaufman, "Calculations of Binary Phase Diagrams", Proc. Intl.
Conf. on Metallurgical Thermochemistry Fundamentals, Brunel
Univ. and NPL, London (1971).
G. A. Alers, J. R. Neighbours, and H. Sata, "Temperature Dependent
Magnetic Contributions to the High Field Elastic Constants of
Nickel and an Fe-Ni Alloy", J. Phys. Chem. Solids 13 40 (1960).
C. H. Shih, B. L. Averbach, and M. Cohen, "Some Characteristics of
the Isothermal Martensitic Transformation", Trans. AIME. 202
183 (1955).
110.
S. R. Pati, Nucleation of the Isothermal Martensitic Transformation
in Iron-Nickel-Manganese Alloys, Sc.D. Thesis, M.I.T. (1967).
111.
E. Owen and Y. Liu, "Further X-Ray Study of the Equilibrium Diagram
of the Iron-Nickel System", J. Iron and Steel Inst. 163 132
(1949).
112. C. D. Graham, Jr., "Magnetic Annealing", Magnetic Properties of
Metals and Alloys, Metals Park, Ohio: American Society for
Metals, p. 288 (1958).
113.
D. A. Colling, "Magnetic Investigation of Martensite-Austenite
Transformations in Fe-Ni-Co Alloys," Met. Trans. 2 2889
(1971).
114. R. J. Weiss and K. J. Tauer, "A Metallurgical Slide Rule for
Determining Magnetic Properties of 3D Transition Alloys",
Theory of Alloy Phases, Metals Park, Ohio: American Society
for Metals, p. 290 (1955).
164
115. M. Hanson, Constitution of Binary Alloys. N.Y.: McGraw-Hill
Book Company, Inc., (1958).
116.
G. F. Boilling, A. Arrott, and R. H. Richman, "A Study of Curie
Temperatures and the Effect of Carbon in Face-Centered Cubic
Iron-Nickel Alloys", phys. stat. sol. 26 743 (1968).
117.
R. L. Fullman, "Measurement of Particle Sizes in Opaque Bodies",
Trans. AIME 197 447 (1953).
165
BIOGRAPHICAL NOTE
The author was born on June 21, 1945 in Cleveland, Ohio. He
graduated from Garfield Heights High School in June, 1962 and received
his B. S. degree from Case Institute of Technology in June, 1966.
His M. S. thesis at the newly federated Case-Western Reserve University
involved studying the Bordoni internal friction peaks in plastically
deformed molybdenenum single crystals.
In July, 1968, he was married
to his wife, Elaine and moved to the Boston/Cambridge area.
In Sep-
tember of that year, he entered graduate school at the Massachusetts
Institute of Technology and has remained there to the present time.
His formal education has been supplemented with summer jobs in the
construction industry and, later at General Electric, N.A.S.A. Lewis
Research Center, and the U. S. Steel Laboratory for Fundamental
Research.
After completing his Sc.D. requirements, he plans to extend
his work with a short postdoctoral appointment at M.I.T. followed by
a N.A.T.O. Postdoctoral Fellowship at Oxford University under Professor
J. W. Christian.
His extracurricular activities include fishing, skiing, and scuba
diving, while his immediate professional objective is university teaching and research in materials science.
Download